TW202002656A - Methods and devices for performing sample adaptive offset (SAO) filtering - Google Patents

Abstract

The present invention relates to methods for improving the coding efficiency of images subjected to the SAO filtering. According to a first aspect of the invention, there is provided a method of performing sample adaptive offset (SAO) filtering on an image comprising a plurality of image parts, the method comprising: obtaining an edge offset to perform a SAO filtering; applying an edge offset classification to at least one pixel of the image using a predetermined sign method for determining an index of the edge offset; wherein the predetermined sign method is signalled in the bitstream. Correspondingly, there is provided a method of providing an edge offset to perform a sample adaptive offset (SAO) filtering on an image, the method comprising: computing an edge offset to perform the SAO filtering; selecting a sign method among a plurality of possible sign methods for determining an index of the edge offset; and signalling the selected sign method in the bitstream thereby allowing the edge offset classification to be performed according to the selected sign method.

Description

Method and device for performing sampling adaptive offset (SAO) filtering

The invention relates to video encoding and decoding.

In recent years, the Joint Video Experts Group (JVET), a collaborative organization formed by MPEG and ITU-T research group 16’sVCEG, has begun work on a new video coding standard called multi-function video coding (VVC). The goal of VVC is to provide a significant improvement in the compression performance of the current HEVC standard (ie, typically twice that of the previous one) and be completed in 2020. The main target applications and services include but are not limited to 360-degree and high dynamic range (HDR) video. In summary, JVET evaluated the responses of 32 organizations that used formal subjective tests from independent test laboratories. When compared to the use of HEVC, some proposals exhibit a typical compression efficiency gain of 40% or more. The special effects are shown on ultra-high resolution (UHD) video test materials. Therefore, we can expect that in the final standard, the compression efficiency gain will far exceed the target by more than 50%.

The JVET exploration model (JEM) uses all HEVC tools. One of these tools is sample adaptive offset (SAO) filtering. However, the SAO in the JEM reference software is less efficient than the HEVC reference software. Compared to other loop filters, this is caused by fewer evaluations and low signal transmission efficiency.

In the document JVET-J0024, it is disclosed to modify the SAO edge offset classification. This modification gave interesting results when testing the low-latency P architecture, but there was a loss for the low-latency B architecture on top of building VTM software for the VVC standard.

Therefore, it is desired to improve the coding efficiency of SAO filtered video.

The present invention has been designed to address one or more of the aforementioned concerns.

According to the first aspect of the present invention, a method for performing sample adaptive offset (SAO) filtering on an image having a plurality of image parts is provided. The method includes: Get the edge offset to perform SAO filtering; Using a predetermined symbol method to determine the index of the edge offset, apply the edge offset classification to at least one pixel of the image; The predetermined symbol method is sent in the bit stream.

Correspondingly, there is provided a method for providing edge offset for performing sample adaptive offset (SAO) filtering on an image. The method includes: Calculate the edge offset to perform SAO filtering; Choose one of the most possible symbologies to determine the index of the edge offset; and In the bit stream, the selected symbol method is sent to allow the edge offset classification to be performed according to the selected symbol method.

Therefore, according to the method of the first aspect of the present invention, it can improve the coding efficiency of the image subjected to SAO filtering.

At the same time, they also provide better flexibility than previous techniques that use the same notation for all edge offsets. Therefore, these methods are better suited to quantization errors and video content.

The selection characteristics of the first aspect are further defined in the ancillary items.

According to an embodiment, the predetermined symbol method is signaled in the bit stream at the CTU level.

The advantages of the CTU level signal edge symbol method are based on the effective adaptation of the frame/slice content and quantization error, which results in an improvement in coding efficiency compared to previous techniques.

According to an embodiment, the predetermined symbol method is signaled in the bit stream in frame, slice, or sequence levels.

The advantage of this embodiment may be that when the frame/slice content and quantization error are relatively homogeneous within the current slice/picture frame, the coding efficiency may be improved. The main advantage of sending messages in a slice file is that the number of possible values for the symbol method may be higher than those used for CTU level signal sending methods.

According to an embodiment, the predetermined symbol method is CABAC coded.

The advantage of this implementation method may be that when the content of the frame/slice and the quantization error are relatively homogeneous within the current slice/frame, the coding efficiency increases. As a result, an edge offset symbol method can be selected for many CTUs.

According to the embodiment, the predetermined symbol method is predicted.

The advantage of the second embodiment and all its sub-embodiments is that due to the spatial correlation of the signal content and the use of similar quantization errors between the CTUs of the current slice/picture frame, the coding efficiency is increased. This is because when the signal of the current slice is similar by CTU and when the quantization error is also similar, the optimal symbol selection method on the encoder side is usually similar.

According to an embodiment, the predetermined symbol method is also selected on the encoder side to minimize the rate-distortion cost.

The advantage of this embodiment may be that it should reach the maximum coding efficiency for slice/picture frame level signaling.

According to the embodiment, the predetermined symbol method on the encoder side is selected according to the parameters of the current slice or picture frame.

According to an embodiment, the predetermined sign method considers the number N corresponding to the limit value to define the sign function.

According to the embodiment, the predetermined sign method is selected on the encoder side according to the discontinuous majority limit value N.

Since continuous values do not correspond to the statistics of natural content and related quantization errors, these embodiments may be interesting when transmitting at the CTU level or lower. One advantage is improved coding efficiency.

According to the embodiment, the predetermined symbol method is selected on the encoder side according to the majority limit value N depending on the slice type.

According to an embodiment, the predetermined symbol method is selected on the encoder side according to a majority limit value N that depends on color components.

The advantage of this embodiment may be due to the improved coding efficiency of the N-value of the adaptation group of each color component type.

According to an embodiment, the predetermined symbol method is selected on the encoder side based on a majority limit value N that only depends on brightness.

The advantage of this embodiment may be the improvement in coding efficiency by removing the symbology selected for chroma components.

According to the embodiment, the predetermined symbol method is selected on the encoder side according to the majority limit value N transmitted in the slice header.

The advantage of this embodiment may be to increase the coding efficiency by setting a useful set for each slice on the encoder side.

According to the embodiment, the predetermined symbol method is selected on the encoder side according to the majority limit value N.

According to the embodiment, the majority limit value N depends on the color component.

According to a second aspect of the present invention, there is provided a method of performing sample adaptive offset (SAO) filtering on an image including a plurality of image parts. The method includes: Obtain most edge offsets to perform SAO filtering; The symbol of at least one edge offset of the majority of edge offsets is transmitted in the bit stream.

According to an embodiment, the method may also include applying an edge offset classification to the pixels of the image, where the edge offset classification takes into account at least one edge offset symbol sent from the majority of edge offsets in the bitstream .

Correspondingly, there is provided a method for providing edge offset to perform sample adaptive offset (SAO) filtering on an image. The method includes: Obtain the majority of edge offsets to perform the SAO filtering; Calculate the index of the edge offset by applying the sign method to a symbol parameter N for each edge offset; and At least one edge offset symbol of the majority edge offset is signaled in the bit stream.

Therefore, the method according to the second aspect of the present invention makes it possible to improve the coding efficiency of the image subjected to SAO filtering.

The optional features of the second aspect are further defined in the ancillary items.

According to an embodiment, the sign of the edge offset is sent for all edge offsets of the majority of edge offsets, where the sign parameter N in the sign function used to determine the index of the edge offset is positive, The sign of this edge shift is strictly positive.

According to an embodiment, if all symbol parameters N in the symbol function used to determine the index of the edge offset are higher than a predetermined threshold, then the symbol of the edge offset is sent to all edge offsets of the majority of edge offsets.

The advantage of this embodiment may be that when the value N used in the symbol method is high, the coding efficiency is increased compared to the implicit symbol signaling used for the offset for edge offset classification.

According to an embodiment, at least one edge offset to which the symbol of the edge offset is signaled belongs to the peak and valley structure.

According to an embodiment, at least one edge-shifted symbol is sent regardless of what predetermined symbol method is used to determine the index of the edge-shift.

According to an embodiment, the symbol of the edge offset only signals the edge offset, where the symbol parameter N used to determine the indexes of these edge offsets is higher than a predetermined threshold.

According to an embodiment, the symbol of the edge offset only signals the edge offset, where the symbol parameter N used to determine the index of these edge offsets is not zero.

According to an embodiment, the edge-shifted symbol only signals the luminance component.

The advantage of this embodiment may be improved coding efficiency compared to the symbol sender for all color components. This is because for chroma components, implicit signaling may be more efficient than explicit signaling of symbols.

According to a third embodiment of the present invention, a method for performing sample adaptive offset (SAO) filtering on an image including a plurality of image parts is provided. The method includes: Obtain most edge offsets to perform SAO filtering; Where at least one edge-offset symbol of the majority of edge offsets is sent in the bitstream; and Obtaining the majority of edge offsets includes predicting at least one edge offset whose symbol is signaled in the bitstream.

Correspondingly, there is provided a method for providing edge offset to perform sample adaptive offset (SAO) filtering on an image. The method includes: Obtain most edge offsets to perform SAO filtering; Obtain the sign of the edge offset for each edge offset; The method further includes signaling at least one edge-shifted symbol of the majority of edge shifts in the bitstream; and Obtaining the majority of edge offsets includes predicting at least one edge offset whose symbol is signaled in the bitstream.

Therefore, the method according to the third aspect of the present invention may improve the coding efficiency of the image subjected to SAO filtering.

The optional features of the third aspect are further defined in the ancillary items.

According to an embodiment, at least one edge offset is predicted from a preset value.

According to an embodiment, the preset value is transmitted in the bit stream.

According to an embodiment, the preset value is predetermined for at least one sequence, for example all sequences.

According to an embodiment, the method further includes setting the symbol of the preset value.

According to an embodiment, the preset value depends on the sign parameter N used in the sign function that determines the index of the edge offset.

The advantage of this embodiment may be an increase in coding efficiency compared to a fixed value for all edge offset sign methods. This is because when the N value of the symbol method increases, the absolute offset value also increases.

According to an embodiment, the method further includes applying an edge offset classification to the pixels of the image to classify the pixels into four types, each type related to a given edge offset, and corresponding to peaks and valleys related to the first and The absolute predictor value for the last type of edge offset is higher than the absolute predictor value for other types of edge offset.

This embodiment may be particularly advantageous because the absolute average of offsets O1 and O4 is generally greater than the absolute average of offsets O2 and O3. Therefore, this embodiment can provide better coding efficiency.

According to an embodiment, the method further includes applying an edge offset classification to the pixels of the image to classify the pixels into four types, each type related to a given edge offset, and corresponding to peaks and valleys related to the first and The absolute predictor value of the last type of edge offset is the same.

According to an embodiment, the absolute predictor values related to other types of edge offsets are the same.

Preferably, the procedure is simplified and matched with the statistical selection of offset values.

According to an embodiment, the edge offset classification takes into account at least one edge offset symbol of most edge offsets, which is signaled in the bit stream.

According to an embodiment, the at least one edge offset is predicted by a symbol parameter N in a symbol function used to determine the index of the edge offset.

According to an embodiment, the symbols in the majority of edge offsets are signaled that at least some edge offsets in the bitstream are not predicted.

According to an embodiment, the method further includes applying an edge offset to the pixels of the image to classify the pixels into four types, each type is associated with a given edge offset, and only the peak and valley An edge offset from the last type is predicted.

The advantage of this embodiment may be increased coding efficiency. This is because the absolute average of the offsets O1 and O4 is larger than the offsets O2 and O3 on average. In addition, the absolute values of O2 and O3 are often lower than 1. As a result, it may not be necessary to predict them and send a symbol.

According to the embodiment, the edge offset is only predicted for the luminance component.

The advantage of this embodiment may be increased coding efficiency. This is because the absolute offset value used for the luma component is usually higher than that used for the chroma component. The prediction of the luminance component may be effective. Therefore, this embodiment can provide a simplified version because the prediction of chroma shift is avoided.

According to an embodiment, the method further includes applying an edge offset classification to the pixels of the image to classify the pixels into four types, each type related to a given edge offset, and only the peaks and valleys related to the first The edge offset from the last type and the edge offset of the luminance component are predicted.

According to an embodiment, at least one edge offset is predicted from adjacent edge offset values.

According to an embodiment, only when the adjacent edge offset is determined using the same sign method, the at least one edge offset is predicted from the adjacent edge offset value.

According to a fourth aspect of the present invention, there is provided a method for performing sample adaptive offset (SAO) filtering on an image including a plurality of image parts. The method includes: Obtain most edge offsets to perform SAO filtering; Wherein obtaining the majority of edge offsets includes predicting at least one edge offset; and The sign of the at least one edge offset is implicit and the minimum value of the edge offset is set to a preset value other than zero.

Correspondingly, there is provided a method for providing edge offset to perform sample adaptive offset (SAO) filtering on an image. The method includes: Calculate the majority of edge offsets to perform SAO filtering; Where calculating most edge offsets includes predicting at least one edge offset, and The sign system of the at least one edge offset is implicit and the minimum value of the edge offset is set to a preset value other than zero.

Therefore, the method according to the fourth aspect of the present invention may improve the coding efficiency of SAO-filtered images.

Since it is not necessary to add the offset symbol as a syntax element, the method according to this aspect can be regarded as a simplified version of the embodiment in which the symbol is transmitted.

The optional feature of the fourth embodiment is further defined in the ancillary items.

According to an embodiment, the at least one edge offset is predicted from a preset value.

According to an embodiment, the preset value is transmitted in the bit stream.

According to an embodiment, the predetermined value is determined for at least one sequence, for example all sequences.

According to an embodiment, the method further includes setting the symbol of the preset value.

According to an embodiment, the preset value depends on the sign parameter N in the sign function used to determine the index of the edge offset.

The advantage of this embodiment may be an increase in coding efficiency compared to a fixed value used for all edge offset sign methods. This is because when the N value of the symbol method increases, the absolute offset value also increases.

According to an embodiment, the method further includes applying an edge offset to the pixels of the image to classify the pixels into four types, each type related to a given edge offset, and corresponding to peaks and valleys related to the first and last The absolute predictor value of the type of edge offset is higher than that of other types of edge offset.

This embodiment can be quite advantageous because the absolute average of offsets O1 and O4 is often greater than the absolute average of offsets O2 and O3. Therefore, this embodiment can provide better coding efficiency.

According to an embodiment, the method further includes applying edge offset classification to the pixels of the image to classify the pixels into four types, each type related to a given edge offset, and the first and last corresponding to peaks and valleys The absolute predictor value of the type-dependent edge offset is the same.

According to an embodiment, the absolute predictor values for other types of edge offsets are the same.

Preferably, the procedure is simplified and adapted to the statistical selection of offset values.

According to an embodiment, the edge offset classification takes into account symbols of at least one edge offset of the majority of edge offsets that are signaled in the bitstream.

According to an embodiment, the at least one edge offset is predicted by a symbol parameter N in a symbol function used to determine the index of the edge offset.

According to an embodiment, at least some edge offsets of the majority of edge offsets are not predicted.

According to an embodiment, the method further includes applying an edge offset classification to the pixels of the image to classify the pixels into four types, each type related to a given edge offset, and only related to the first corresponding to peaks and valleys The edge offset from the last type is predicted.

The advantages of this embodiment can increase coding efficiency. This is because the absolute average of the offsets O1 and O4 is larger than the offsets O2 and O3 on average. In addition, the absolute values used for O2 and O3 are often below 1. As a result, it may not be necessary to predict them.

According to the embodiment, the edge offset is only predicted for the luminance component.

The advantages of this embodiment can increase coding efficiency. This is because the absolute offset value for luminance is usually higher than the absolute offset value for chroma components. The prediction of the luminance component may be sufficiently effective. Since the prediction of chroma shift is avoided, this embodiment can therefore provide a simplified version.

According to an embodiment, the method further includes applying an edge offset classification to the pixels of the image to classify the pixels into four types, each type is related to a given edge offset, and only the peaks and valleys related to the first The edge offset from the last type and the edge offset of the luminance component are predicted.

According to an embodiment, at least one edge offset is predicted from adjacent edge offset values.

According to an embodiment, only when the adjacent edge offset is determined using the same sign method, the at least one edge offset is predicted from the adjacent edge offset value.

According to another aspect of the present invention, there is provided a method for encoding an image, which includes using one of the foregoing methods to perform sample adaptive offset (SAO) filtering.

According to another aspect of the present invention, a method for decoding an image is provided, which includes using one of the foregoing methods to perform sample adaptive offset (SAO) filtering.

According to another aspect of the present invention, there is provided an apparatus for performing sample adaptive offset (SAO) filtering on an image including a plurality of image parts, the apparatus including a microprocessor configured to perform the steps of one of the aforementioned methods .

According to another aspect of the present invention, there is provided an encoder including the aforementioned device.

According to another aspect of the present invention, there is provided a decoder including the aforementioned device.

According to another aspect of the present invention, there is provided a system including the aforementioned encoder and the aforementioned decoder, configured to communicate via a communication network.

The other aspects of the present invention have similar characteristics and advantages as the aforementioned first, second, third, and fourth aspects.

Because the present invention can be implemented in software, the present invention can be implemented as a computer-readable code for providing to a programmable device on any suitable carrier media. The carrier media is more specifically a tangible carrier media or an appropriate temporary media. State media. The tangible carrier medium may include storage media such as a floppy disk, a CD-ROM, a hard disk drive, a tape device, a solid state memory device, or the like. The transient carrier medium may contain signals such as electrical signals, electronic signals, optical signals, sound signals, magnetic signals, or electromagnetic signals such as microwave or RF signals.

Figure 1 relates to the coding structure used in the High Efficiency Video Coding (HEVC) video standard. The video sequence 1 consists of a continuous digital image i. Each such digital image is represented by one or more matrices. The matrix coefficients represent pixels.

The image 2 of this sequence can be cut into slices 3. In some examples, slices can constitute the entire image. These slices are cut into non-overlapping coding tree units (CTU). The coding tree unit (CTU) is the basic processing unit of the high efficiency video coding (HEVC) video standard and the conceptual structure corresponds to the macroblock unit used in several previous video standards. The CTU is also sometimes referred to as the largest coding unit (LCU). The CTU has luminance and chrominance component parts, and each component part is called a coding tree block (CTB). These different color components are not shown in Figure 1.

The CTU is usually 64 pixels by 64 pixels. Each CTU can then be split into smaller variable size coding units (CUs) 5 using quaternary tree decomposition and iteration.

The coding unit is a basic coding element and is composed of two subunits called a prediction unit (PU) and a conversion unit (TU). The maximum size of PU or TU is equal to CU size. The prediction unit corresponds to the partition of the CU used to predict the pixel value. Various partitions from CU to PU are possible as shown in 606, including partitioning into 4 square PUs and 2 different partitions into 2 rectangular PUs. The conversion unit is a basic unit subjected to spatial conversion using DCT. CUs can be partitioned into TUs according to quaternary tree representation 607.

Each slice is embedded in a network abstraction layer (NAL) unit. In addition, the coding parameters of the video sequence are stored in dedicated NAL units called parameter sets. In HEVC and H.264/AVC, two parameter group NAL units are used: first, a sequence parameter group (SPS) NAL unit, which collects all parameters that have not changed throughout the video sequence. Typically, it handles encoding profiles, the size of video frames, and other parameters. Second, the Picture Parameter Set (PPS) NAL unit, which contains parameters that change from one image (or picture frame) to another in the sequence. HEVC also contains Video Parameter Set (VPS) NAL units, which contain parameters describing the entire structure of the bitstream. VPS is a new type of parameter set defined in HEVC and applied to all layers of the bitstream. One layer can contain most time sublayers, and all versions of the 1-bit stream are restricted to a single layer. HEVC has certain layering extensions for zooming and multi-view viewing, and these will make multiple layers with a backward compatible version 1 base layer.

FIG. 2 illustrates a data communication system in which one or more embodiments of the present invention can be implemented. The data communication system includes a transmission device, in this case a server 201, which is operable to transmit data packets of the data stream through the data communication network 200 to the receiving device, which in this example is the client terminal 202. The data communication network 200 may be a wide area network (WAN) or a local area network (LAN). This network can be, for example, a wireless network (Wifi/802.11a or b or g), an Ethernet network, an Internet network or a hybrid network composed of several different networks. In a specific embodiment of the present invention, the data communication system may be a digital television broadcasting system, in which the server 201 sends the same data content to multiple clients.

The data stream 204 provided by the server 201 may be composed of multimedia data representing video and audio data. In some embodiments of the invention, the audio and video data streams can be captured by the server 201 using a microphone or a camera, respectively. In some embodiments, the data stream may be stored in the server 201 or received from the server 201 by another data provider or generated at the server 201. The server 201 is provided with an encoder for encoding video and audio streams, more specifically, a compressed bit stream for transmission, which is a reduced representation of the data presented to the encoder as input.

In order to obtain a better ratio of the quality of the transmitted data to the amount of the transmitted data, the compression of the video data may be based on the HEVC format or H.264/AVC format, for example.

The client 202 receives the transmitted bit stream and decodes the reconstructed bit stream to play video images on the display device and audio data on the speaker.

Although the streaming scenario is considered in the example of FIG. 2, it will be appreciated that in some embodiments of the invention, data communication between the encoder and the decoder may be performed using media storage devices such as optical discs, for example.

In one or more embodiments of the invention, the video image is transmitted with data representing offset compensation to be applied to the reconstructed pixels of the image to provide filtered pixels in the final image.

FIG. 3 illustrates a processing device 300 configured to implement at least one embodiment of the present invention. The processing device 300 may be a device such as a microcomputer, a workstation, or a lightweight portable device. The device 300 includes a communication bus 313 connected to: -A central processing unit 311, such as a microprocessor labeled CPU; -Read-only memory 307 marked as ROM for storing computer programs to implement the present invention; -Random access memory 312 marked as RAM, used to store executable code of the method of the embodiment of the present invention, and a register suitable for recording variables and parameters required for the implementation of the method, the method including implementation according to the present invention Examples of methods for encoding sequences of digital images and/or decoding bit streams; and -The communication interface 302 connected to the communication network 303, on which the processed digital data can be transmitted or received.

Optionally, the device 300 may also contain the following components: -Data storage means 304, such as a hard disk, for storing computer programs for implementing the method of one or more embodiments of the present invention and data used or generated during the execution of one or more embodiments of the present invention; -A disk drive 305 for the disk 306, which is suitable for reading data from the disk 306 or writing data to the disk; -The screen 309 uses the keyboard 310 or other pointing methods to display data and/or services as a graphical interface with the user.

The device 300 may be connected to various peripherals, such as a digital camera 320 or a microphone 308, each connected to an input/output card (not shown) to supply the multimedia material to the device 300.

The communication bus provides communication and interoperability between various components included in the device 300 or connected thereto. The representative of the bus is not limitative. In particular, the central processing unit is operable to send commands to any element of the device 300 directly or through another element of the device 300.

The disc 306 can be replaced by any information medium, for example, a CD-ROM, a writable disc or not, a ZIP disc or a memory card, and, in general terms, by means of information storage, it can be a microcomputer or Those read by the microprocessor, integrated or not integrated into the device, may be removable and suitable for storing one or more programs that perform methods and/or decoding bits that encode sequences of digital images according to the present invention The streaming method can be executed.

The executable code may be stored in the read-only memory 306, or on the hard disk 304 or on the removable digital media, such as the disk 306 as described above. According to a variation, the executable code of the program may be received by the communication network 303 via the interface 302 and stored in one of the storage means of the device 300, such as the hard disk 304, before being executed.

The central processing unit 311 is adapted to control and direct the execution of instructions or the execution of a program or part of the software code of most programs in accordance with the present invention. The instructions are stored in one of the aforementioned storage means. At startup, programs or most programs stored in non-volatile memory, such as hard disk 304 or read-only memory 306, are transferred into random access memory 312 and used to store variables and In the parameter register, the random access memory 312 then contains the executable code of the program or most programs.

In this embodiment, the device is a programmable device that uses software to implement the present invention. However, alternatively, the present invention may be implemented in hardware (for example, in the form of an application specific integrated circuit or ASIC).

FIG. 4 illustrates a block diagram of an encoder according to at least one embodiment of the present invention. The encoder is represented by a connection module, and each module is suitable for implementing at least one corresponding step of a method in the form of planning instructions executed by the CPU 311 of the device 300, for example, according to one or more implementations of the present invention. For example, at least one embodiment implementing an image encoding a sequence of images.

The original sequence digital images i0 to in 401 are received by the encoder 400 as input. Each digital image is represented by a set of samples called pixels. Recall that a pixel contains three components and a sample indicates one of them.

After the encoding process is implemented, the bit stream 410 is output by the encoder 400. The bit stream 410 includes a plurality of coding units or slices. Each slice includes a slice file header and a slice body. The slice file header is used to transmit code values for coding parameters of the slice. The slice body includes coded video data.

The input digital images i0 to in 401 are cut into pixel blocks by the module 402. The blocks correspond to the image part and can be of variable size (for example, 4×4, 8×8, 16×16, 32×32, 64×64, 128×128 pixels and several rectangular block sizes can also be considered ). The encoding mode of each input block can be selected. Two series of coding modes are provided: a coding mode based on spatial prediction coding (intra prediction), and a coding mode based on temporal prediction (inter coding, merge, skip (SKIP)). Maybe the coding mode is tested.

The module 403 implements an intra-prediction process, in which a given block to be encoded is predicted by a predictor calculated by neighboring pixels of the block to be encoded. The indication of the selected intra-predictor and the difference between the given block and its predictor are coded to provide a residual value if the intra-coding is selected.

The time prediction is implemented by the motion estimation module 404 and the motion compensation module 405. First, a reference image/picture is selected from a set of reference images/pictures 416, and a portion of the reference image/picture that is also referred to as a reference area or image portion that is closest to the given block to be encoded It is selected by the motion estimation module 404. The motion compensation module 405 then uses the selected region to predict the block to be encoded. The difference between the selected reference area, also called the residual block, and the given block is calculated by the motion compensation module 405. The selected reference area is indicated by the motion vector.

Therefore, in both cases (spatial and temporal prediction), the residual value is calculated by subtracting the original block from the prediction.

In the intra prediction implemented for the module 403, the prediction direction is coded. In temporal prediction, at least one motion vector is encoded.

If inter prediction is selected, the information about the motion vector and the residual block is coded. In order to further reduce the bit rate, assuming that the actions are homogeneous, the action vector is encoded by the difference about the action vector predictor. The motion vector predictor of a set of motion information predictors is obtained from the motion vector field 418 by the motion vector prediction and encoding module 417.

The encoder 400 further includes a selection module 406 for selecting an encoding mode by applying an encoding cost criterion such as a rate distortion criterion. In order to further reduce redundancy, (for example, DCT) conversion is applied to the residual block by the conversion module 407, and then the obtained conversion data is quantized by the quantization module 408 and entropy encoded by the entropy encoding module 409. Finally, the encoded residual block of the encoded current block is inserted into the bit stream 410.

The encoder 400 also performs decoding of the encoded video to generate a reference video for subsequent motion estimation of the video. This allows the encoder and decoder receiving the bitstream to have the same reference image. The dequantization module 411 performs dequantization of the quantized data, and then is inversely converted by the inverse conversion module 412. The intra prediction module 413 uses prediction information to determine which predictor is used for a given block and the motion compensation module 414 actually adds the residual value obtained by the module 412 to the reference obtained by the set of reference images/pictures 416 area.

Post-filtering is then applied by module 415 to filter the reconstructed frame of pixels. In an embodiment of the present invention, a SAO loop filter is used, in which a compensation offset is applied to the pixel values of the reconstructed pixels of the reconstructed image.

FIG. 5 is a flowchart of the steps of a loop filtering process according to at least one embodiment of the invention. In the initial step 51, the encoder generates a full frame reconstruction. Furthermore, in step 52, the deblocking filter is applied to this first reconstructed picture frame to generate a deblocked reconstructed picture frame in step 53. The goal of the deblocking filter is to remove block artifacts generated for residual quantization and block motion compensation or intra-block prediction. These artifacts are visible and important at low bit rates. The deblocking filter operates to smooth the boundary of the block according to the characteristics of two adjacent blocks. The coding mode of each block, the quantization parameter used for residual coding, and the adjacent pixel difference at the boundary are considered. The same criteria/category is applied to all frames, and no additional information needs to be sent. The deblocking filter improves the visual quality of the current frame by removing block artifacts, which also improves motion estimation and motion compensation for subsequent frames. Indeed, the high frequencies of the block artifacts are removed, and, therefore, these high frequencies do not require residual compensation for the texture of subsequent frames.

After the deblocking filter, the deblocking reconstruction frame is filtered in step 54 by using the SAO parameters determined according to the embodiment of the invention for the sample adaptive offset (SAO) loop filter. The resulting picture frame 55 can then be filtered with adaptive loop filter (ALF) in step 56 to produce a reconstructed picture frame 57 which will be displayed and used for the following picture frame as a reference picture frame.

In step 54, each pixel of the frame area is classified into a class or group. The same offset value is added to each pixel value belonging to a certain class or group.

The derivation of SAO parameters of the sample adaptive offset filtering in different embodiments of the present invention will be explained in more detail with reference to FIGS. 12 to 38 as follows.

FIG. 6 illustrates a block diagram of a decoder 60. The decoder can be used to receive data from an encoder according to an embodiment of the present invention. The decoder is represented by connected modules, and each module is adapted to implement the corresponding method steps implemented by the decoder 60, for example, in the form of program instructions executed by the CPU 311 of the device 300.

The decoder 60 receives a bit stream 61 including coding units, and each coding unit is composed of a header including information of coding parameters and a body including coding video data. As explained with reference to FIG. 4, the encoded video data is entropy encoded, and for a given block, the motion vector predictor index is encoded on a predetermined number of bits. The received encoded video data is entropy decoded by the module 62. The residual data is then dequantized by the module 63, and then inversely converted by the module 64 to obtain pixel values.

The mode data indicating the encoding mode is also entropy decoded, and according to the mode, an INTRA type decoding or an INTER type decoding is performed on the encoding block of the image data.

In the intra mode, the intra prediction sub-system is determined by the intra prediction module 65 according to the intra prediction mode specified in the bit stream.

If the mode is inter, the motion prediction information is extracted from the bit stream to find the reference area used by the encoder. The motion prediction information is composed of the reference frame index and the residual value of the motion vector. The motion vector predictor is added to the residual value of the motion vector to be the motion vector obtained by the motion vector decoding module 70.

The motion vector decoding module 70 applies motion vector decoding to each current block encoded for motion prediction. Once the index of the motion vector predictor for the current block has been obtained, the actual value of the motion vector related to the current block can be decoded and used by the module 66 to perform motion compensation. The reference video part indicated for decoding the motion vector is extracted from the reference video 68 to perform motion compensation 66. The motion vector column data 71 is updated with the decoded motion vector to be used for prediction of the subsequent decoded motion vector.

Finally, get the decoding block. The post filter is applied by the post filter module 67, which is similar to the filter module 815 after the encoder is applied as described with reference to FIG. The decoded video signal 69 is finally provided by the decoder 60.

Compared to deblocking filters where no information is transmitted, the goal of SAO filtering is to improve the quality of the reconstructed frame by sending additional data in the bitstream. As described above, each pixel is classified into a predetermined class or group and the same offset value is added to each pixel sample of the same class/group. For each class, an offset is encoded in the bitstream. SAO loop filtering has two SAO types: edge offset (EO) type and band offset (BO) type. Examples of the edge shift type are schematically illustrated in FIGS. 7A and 7B, and examples of the band shift type are schematically illustrated in FIG. 8.

In HEVC, SAO filtering is applied on a CTU-by-CTU basis. In this case, the parameters required to perform SAO filtering (SAO parameter set) are selected for each CTU on the encoder side and on the decoder side, for each CTU, the required parameters are decoded and/or derived. This provides the possibility to easily encode and decode video sequences by processing each CTU immediately without causing a delay in the entire frame processing. Furthermore, when SAO filtering is enabled: according to the relevant parameters transmitted in the bitstream for each classification, only one SAO type of edge offset filter or band offset filter is used. One of the SAO parameters in HEVC is the SAO type parameter sao_type_idx, which indicates that for this CTU, the EO type, BO type, or SAO-free filtering system is selected for the relevant CTU.

SAO parameters for a given CTU can be copied from the upper or left CTU, for example, without transmitting all SAO data. One of the SAO parameters in HEVC is the sao_merge_up flag. When it is set, the SAO parameters representing the subject CTU (except the sao_merge_up flag) should be copied by the upper CTU. The other of the SAO parameters in HEVC is the sao_merge_left flag. When it is set, it indicates that the SAO parameters used for the subject CTU should be copied by the left CTU.

SAO filtering can be individually applied to different color components of the picture frame (for example, YUV). For example, one set of SAO parameters may be provided for the luma component Y and another set of SAO parameters may be provided jointly for both the chroma components U and V. At the same time, within the set of SAO parameters, one or more SAO parameters can also be used as a common filtering parameter for two or more color components, while other SAO parameters are dedicated (per component) filtering for color components parameter. For example, in HEVC, the SAO type parameter sao_type_idx is shared by U and V, and the same is true for EO type parameters, which represent a type used for EO filtering (see below), and BO type parameters, which represent the type used for BO filtering A group of classified (per component) SAO parameters for U and V.

As shown in FIG. 10, when the SAO type is an edge type (603), the sign of the offset is implicit as shown in FIG. 7B. Therefore, the symbol is not read (610). This implicit symbol decision is adopted when HEVC is standardized, because the symbol transmission does not improve the coding efficiency of SAO. In the current VTM software, if the implicit symbols are replaced by explicit symbols, as in the original version of HEVC, you can see a reduction in coding efficiency.

A description of the edge shift type in HEVC is now provided with reference to FIGS. 7A and 7B.

The edge shift type involves determining the edge index for each pixel by comparing its pixel value with the values of two adjacent pixels. Furthermore, the two adjacent pixels depend on a parameter, which indicates the direction of the two adjacent pixels relative to the current pixel. These directions are 0 degrees (horizontal direction), 45 degrees (diagonal direction), 90 degrees (vertical direction) and 135 degrees (second diagonal direction). These four directions are illustrated in Figure 7A.

The table of FIG. 7B gives the offset value of the pixel value to be applied to the specific pixel "C" according to the values of two adjacent pixels Cn1 and Cn2 on the decoder side.

When the value of C is lower than the two values of the adjacent pixels Cn1 and Cn2, the offset of the pixel value added to the pixel C is "+O1". When the pixel value of C is lower than a pixel value of its neighboring pixel (Cn1 or Cn2) and C is equal to a value of its neighboring pixel, the offset added to the sampling value of this pixel is "+O2".

When the pixel value of c is lower than one of the pixel values of the neighboring pixels (Cn1 or Cn2) and the pixel value of C is equal to a value of its neighboring pixels, the offset applied to the pixel sampling is "-O3". When the value of C is greater than the two values of Cn1 or Cn2, the offset applied to this pixel sample is "-O4".

When none of the above states match the current sample and its neighboring pixels, no offset value is added to the current pixel C, as depicted by the edge index value "2" of the table.

It is important to note that in the specific case for the edge offset type, the absolute value of each offset (O1, O2, O3, O4) is encoded in the bitstream. The symbol to be applied to each offset depends on the edge index to which the current pixel belongs (or the edge index in the HEVC specification). According to the table shown in FIG. 7B, for edge index 0 and for edge index 1 (O1, O2), a positive offset is applied. For edge index 3 and edge index 4 (O3, O4), the negative offset is applied to the current pixel.

In the HEVC specification, the direction of the edge shift among the four directions in FIG. 7A is indicated in the bit stream, and the "sao_eo_class_luma" column indicates the luminance component and the "sao_se_class_chroma" column indicates the chrominance components U and V.

The SAO edge index corresponding to the index value is obtained by the following formula: EdgeIndex=sign(C-Cn2)-sign(Cn1-C)+2 The definition of the function sign(.) is given by the following relationship: sign(x)=1, when x＞0 sign(x)=-1, when x<0, sign(x)=0, when x=0.

To simplify the determination of the edge offset for each pixel, the difference between the pixel value of C and the pixel values of two adjacent pixels Cn1 and Cn2 can be shared by the current pixel C and its neighbors. Indeed, when the SAO edge offset filter is applied using the optical domain scanning order of the pixels of the current CTU or frame, the term symbol (Cn1-C) has been calculated for the previous pixel (more precisely, when the current pixel C' When the current adjacent pixel Cn1 and the adjacent pixel Cn2' are the current active pixel C at this time, it is calculated as C'-Cn2' at a time. As a result, this symbol (cn1-c) does not need to be calculated again.

An explanation of the offset type will now be provided with reference to FIG. 8.

The band offset type in SAO also depends on the pixel value of the sample being processed. One type of SAO band offset is defined as a range of pixel values. Traditionally, for all pixels in a range, the same offset is applied to the pixel value. In the HEVC specification, for each reconstruction block or picture frame area (CTU) of a pixel, the number of offsets used for the band offset filter is four, as illustrated in FIG. 8.

An implementation of SAO band offset subdivides the full range of pixel values into 32 ranges of the same size. These 32 ranges are SAO band offset bands (or classes). The minimum value of the pixel values in this range is system 0 and the maximum value depends on the bit depth of the pixel value according to the following relationship Max=2 ^{bit depth} -1. The full interval that classifies these pixels into 32 ranges contains a 5-bit check, which requires pixel values that are classified for fast implementation, that is, only the first 5 bits (5 most efficient bits) are checked to Classify pixels into one of 32 categories/ranges of the full range.

For example, when the bit depth is 8 bits per pixel, the maximum value of one pixel may be 255. Therefore, the range of pixel values is between 0 and 255. For this 8-bit bit depth, each band or class contains 8 pixel values.

According to the HEVC specification, a group of 40 bands represented by the gray area (40) is used, the group has four consecutive bands 41, 42, 43, and 44, and the information system is sent in the bit stream to indicate the The position of the group, for example, the position of the first band of the four bands. The syntax element at this position is indicated in the "sao_band_position" column in the HEVC specification. This corresponds to the start of band 41 in FIG. 8. According to the HEVC specification, four offsets corresponding to the four bands are signaled in the bit stream.

FIG. 9 is a flowchart showing the procedure steps for decoding SAO parameters according to the HEVC specification. The procedure of FIG. 9 is applied to each CTU to generate a set of SAO parameters for all components. In order to avoid coding a set of SAO parameters per CTU (very expensive), the prediction scheme is used in the CTU mode. This prediction mode involves checking whether the CTU on the left side of the current CTU uses the same SAO parameters (this is indicated in the bitstream by a flag called "sao_merge_left_flag"). If not, a second check is performed on the CTU on the current CTU (this is indicated in the bitstream by a flag called "sao_merge_up_flag"). This prediction technique reduces the amount of data representing SAO parameters used in CTU mode. The steps of the program are described below.

In step 503, "sao_merge_left_flag" is read and decoded by the bitstream 502. If its value is true, the procedure proceeds to step 504, where the SAO parameters for the left CTU are copied for the current CTU. This allows the type of YUV used for the SAO filter of the current CTU to be determined in step 508.

If in step 503 the result is no, then "Sao_merge_up_flag" is read out from the bit stream and decoded. If its value is true, the procedure proceeds to step 505, where the SAO parameters of the upper CTU are copied for the current CTU. This causes the type of SAO filter used for the current CTU to be determined in step 508.

If the result is no in step 505, the SAO parameters of the current CTU are read and decoded by the bit stream in step 507, and are used for the luminance Y component and the two U and V components (501) (551) type. The chromaticity shift is independent.

The details of this step will be described later with reference to FIG. After this step, the parameters are obtained and the type of SAO filter is determined in step 508.

In the subsequent step 511, a check is performed whether the three color components (Y and U&V) of the current CTU have been processed. If the result is yes, it is determined that the SAO parameters for the three components have been completed, and the next CTU may be processed in step 510. If not, (only Y is processed) U and V are processed together and the procedure restarts from the initial step 512 described above.

FIG. 10 is a flowchart showing the procedure of parsing the SAO parameters in the bit stream 601 on the decoder side. In the initial step 602, The "sao_type_idx_X" syntax element is read and decoded. The codeword representing this syntax element may use a fixed-length code or may use any arithmetic coding method. The syntax element sao_type_idx_X causes the decision of the type of SAO for the picture frame area to be processed for both the color component Y or the chroma component U&V. For example, for the YUV 4:2:0 sequence, two components are considered: one for Y and one for U and V. "Sao_type_idx_X" can take 3 values depending on the SAO type encoded in the bit stream as follows. '0' corresponds to no SAO, '1' corresponds to the band offset case shown in Fig. 8, and '2' corresponds to the edge offset type filter shown in Figs. 3A and 3B.

In addition, although YUV color components are used for HEVC (sometimes referred to as Y, Cr, and Cb components), it will be understood that other color components, such as RGB color components, can be used in other video encoding schemes. The technique of the present invention is not limited to use with YUV color components, and may be used with RGB color components or any other color components.

In the same step 602, a test is performed to determine whether "sao_type_idx_X" is strictly positive. If "sao_type_idx_X" is equal to "0", it means that if X is set to Y, there is no SAO for this frame area (CTU) of Y, and, if X is set to U and V, then this picture for U and V There is no SAO in the box area. The determination of the SAO parameters is completed and the process proceeds to step 608. Otherwise, if "sao_type_idx" is strictly positive, this means that there is a SAO parameter for this CTU in the bitstream.

Then, the process proceeds to step 606, where the loop system is executed four iterations. The fourth iteration is performed in step 607, where the absolute value of the offset j is read and decoded from the bit stream. These four offsets correspond to the four absolute values of the four edge index offsets (O1, O2, O3, and O4) of the SAO edge offset (see FIG. 7B) or are related to the SAO band offset (see FIG. 8) Four absolute values of the offset of the four ranges.

其中＜＜為左(位元)移位運算子。Note that for the encoding of the SAO offset, the first part is transmitted in the bit stream corresponding to the absolute value of the offset. This absolute value is encoded with a unary code. The maximum value used for the absolute value is given by the following formula:

Where << is the left (bit) shift operator.

This formula means that for 8-bit pixel value bit depth, the maximum absolute value of the offset is 7, and for 10-bit and above pixel value bit depth is 31.

The current HEVC standard modification for extended bit depth video sequences provides a similar formula for pixel values with a bit depth of 12 bits and above. The decoded absolute value may be a quantized value, which is dequantized before being applied to the pixel value at the decoder for SAO filtering. The indication of the use of this quantization is transmitted in the slice header.

For the edge-offset type, only the absolute value is transmitted because the symbol can be inferred as described above.

For the offset type, if the absolute value of the offset is not equal to 0, the symbol is sent in the bit stream as the second part of the offset. When CABAC is used, the sign bit is skipped.

After step 607, the process proceeds to step 603, where a test is performed to determine whether the type of SAO corresponds to the band offset type (sao_type_idx_X==1).

If the result is positive, the symbols used for the offset with offset mode will be decoded in steps 609 and 610 to be read in the bit stream before the following steps 604 are executed unless each offset has a value of zero. Take and decode the position "aso_band_position_X" of the SAO band as shown in FIG.

If in step 603, the result is negative ("Aso_type_idx_X" is set equal to 2), which means that the edge offset type is used. Therefore, the edge offset class (corresponding to directions 0, 45, 90, and 135 degrees) is extracted from the bit stream 601 in step 605. If X is equal to Y, the read syntax element is "sao_eo_class_luma" and if X is set to U and V, the read syntax element is "sao_eo_class_chroma".

When the four offsets have been decoded, the reading of the SAO parameters is completed and the process proceeds to step 608.

FIG. 11 is a flowchart of how to perform SAO filtering on the video part according to the HEVC specification, for example, during step 67 of FIG. 6. In HEVC, this video part is CTU. The same procedure 700 is also applied to the decoding loop of the encoder (step 415 in FIG. 4) to generate reference frames for motion evaluation and compensation of the following frames. This program is related to SAO filtering of a color component (hence, the subscript "_X" in the syntax element has been omitted below).

The initial step 701 includes determining the SAO filtering parameters according to the procedures described in FIGS. 9 and 10. The SAO filtering parameters are determined by the encoder and the encoded SAO parameters are included in the bitstream. Therefore, in step 701, on the decoder side, the decoder reads and decodes the parameters from the bitstream. Step 701 obtains sao_type_idx and if it is equal to 1, step 701 also obtains sao_band_position 702 and if it is equal to 2, step 701 also obtains sao_eo_class_luma (depending on the color component being processed) or sao_eo_class_chroma. If the element sao_type_idx is equal to 0, no SAO filtering is applied. Step 701 also obtains the offset table 703 for the four offsets.

The variable i used to continuously consider each pixel Pi of the current block or frame area (CTU) is set to 0 in step 704. Incidentally, "frame area" and "image area" are used interchangeably in this manual. In this example, the frame area is the CTU in the HEVC standard. In step 706, the pixel P _i is extracted from the frame area 705 containing N pixels. This pixel P _i is classified in step 707 according to the edge offset classification described with reference to FIGS. 7A and 7B or the band offset classification described with reference to FIG. 8. The decision module 708 tests whether _Pi is to be filtered using traditional SAO filtering.

712。此濾波像素值

在步驟713中被插入濾波畫框區域716中。If P _i is a filtering class, the number of related classes j is identified and the related offset value Offset _j is extracted from the offset table 703 in step 710. In traditional SAO filtering, this Offset _{j is} then added to the pixel value P _i in step 711 to generate a filtered pixel value

712. This filtered pixel value

In step 713, it is inserted into the filtered frame area 716.

If P _{i is} not in the category to be filtered by SAO, then P _i (709) is inserted into the filtered frame area 716 in step 713 without filtering.

After step 713, the variable i is incremented in step 714 to filter subsequent pixels of the current frame area 705 (if any-test 715). In step 715, after all pixels have been processed (i>=N), the filtered frame area 716 is reconstructed and can be added to the SAO reconstruction frame (see frame 68 in FIG. 6 or frame 416 in FIG. 4) ).

As mentioned above, the JVET exploration model (JEM) for the future VVC standard uses all HEVC tools. One of these tools is sample adaptive offset (SAO) filtering. However, the SAO in the JEM reference software is less efficient than the HEVC reference software. This results in fewer evaluations and lower signal transmission efficiency compared to other loop filters.

The embodiments of the present invention described below are intended to improve the coding efficiency of SAO by using various techniques to derive the image in the current image from one or more SAO parameters of the parallel image portion in the reference image One or more SAO parameters. These techniques can be referred to as time derivation techniques for SAO parameters. The further embodiments described below are intended to improve the coding efficiency of SAO by using various techniques to derive one or more SAO parameters of one image part of an image from another image part of the same image. Or more SAO parameters. These techniques can be referred to as spatial derivation techniques for SAO parameters.

FIG. 12 is a flowchart illustrating the steps performed by the encoder to determine the SAO parameters for a group (frame or slice) of CTUs. The procedure starts with the current CTU (1101). First, the statistics of all possible SAO types and classes are accumulated in the variable CTUStats (1102). The procedure of step 1102 will be described with reference to FIG. 13 as follows. According to the value group in the variable CTUStats, if the left CTU is in the current slice, the RD cost of the SAO left merge is evaluated (1103), otherwise, the RD cost for the SAO merge is evaluated (1104). Due to the statistics in CTUStats (1102), the new SAO parameters were evaluated for the luma component (1105) and the dichromatic component (1109). (Two chroma components, because in the HEVC standard, chroma components share the same SAO type). For each SAO type (1106), the best RD offset and other parameters for band offset classification are obtained (1107). Steps 1107 and 1110 are explained below with reference to FIGS. 14 and 15 for edge and band classification, respectively. Due to its individual SAO parameters, all RD costs are calculated (1108). For the two chroma components, the best RD offset and parameters are selected in the same way (1111). All this RD cost is compared to select the best SAO parameter set (1115). These RD costs are also compared to independently remove SAO for luminance and chrominance components (1113, 1114). The use of the new SAO parameter set (1115) is compared with the SAO parameter set "merging" or sharing (1116) from the left and upper CTU.

FIG. 13 is a flowchart illustrating steps of an example in which statistics calculated on the encoder side can be applied to an edge offset type filter when conventional SAO filtering is performed. A similar method can also be used for band offset filters.

FIG. 13 illustrates the setting of the variable CTUStats that contains all the information necessary to derive each optimal rate distortion offset for each class. Furthermore, it illustrates the selection of the best SAO parameter set for the current CTU. For each color component Y, U, V (or RGB) (811), each SAO type is evaluated. For each SAO type (812), the variables Sum _j and SumNbPix _j are set to zero in the initial step 801. The current picture frame area 803 contains N pixels. j is the current range number used to determine the four offsets (there are four edge indexes shown in FIG. 7B for the edge offset type or 32 for the pixel values shown in FIG. 8 with the offset type range). Sum _j is the sum of the differences between the pixels in the range j and their original pixels. SumNbPix _j is the number of pixels in the frame area and their pixel values belong to the range j.

In step 802, the variable i used to continuously consider each pixel Pi of the current frame area is set to zero. Then, the first pixel P _{i of the} picture frame area 803 is extracted in step 804. In step 805, the type of current pixel is determined by checking the state defined in FIG. 7B. Then, in step 805, a test is performed. During step 805, a check is performed to determine the class of the pixel values P _i value corresponds to FIG. 7B "None of the above."

If the result is positive, the value "i" is incremented in step 808 to consider subsequent pixels of the frame area 803.

間的差被加至Sum_j 。在下一步驟808中，變數i被增量，以考量畫框區域803的後續像素。Otherwise, if the result is negative in step 806, the next step is 807, where the related SumNbPix _j (ie, the sum of the number of pixels of this class is determined in step 805) is incremented and the original value of P _i

The difference between is added to Sum _j . In the next step 808, the variable i is incremented to consider the subsequent pixels of the frame area 803.

Then, a test is performed to determine whether all pixels have been considered and classified. If the result is negative, the program loops back to step 804 above. Otherwise, if the result is positive, the process proceeds to step 810, where the variables CTUStats for the current color component X and the SAO type SAO_type and the current class j are set equal to Sum _j for the first value and equal to Sum _j for the second value SumNbPix _j . These variables can be used to calculate the optimal offset parameter Offset _j for each class j, for example. This offset Offset _j may be the average of the difference between pixels of class j and their original values. Therefore, Offset _j is given by the following formula:

Note that Offset _j is an integer value. Therefore, the bit rate defined in this formula can be rounded to the nearest value or use the ceiling or floor function.

In terms of distortion, each offset Offset _j is the optimal offset Ooptj.

To evaluate the RD cost for combining SAO parameters, the encoder uses the statistical set in the table CTUCStats. According to the type used for SAO left and considering the type used for Luma Left_Type_Y and the four related offsets O_Left_0, O_Left_1, O_Left_2, O_Left_3, the distortion can be obtained by the following formula:

The variable Shift system is designed for distortion adjustment. When SAO is post-filtering, the distortion should be negative.

The same calculation is applied to the chroma component. The rate distortion cost λ (Lambda) is fixed for these three components. For SAO parameters incorporating a left CTU, the code rate is only 1 flag, which is CABAC encoded.

The coding procedure exemplified in FIG. 14 is applied to find the best offset expressed in terms of rate distortion criterion, and the offset is called ORDj. This procedure is applied in steps 1109 to 1112.

其中λ為拉格朗日(Lagrange)參數及R(Oj)為一函數，其提供有關Oj的碼字元所需的位元數目。In the initial step 901 of the encoding procedure of FIG. 14, the rate distortion value Jj is initialized to the maximum possible value. Then, a loop of Oj from Ooptj to 0 is applied in step 902. Note that Oj is modified by 1 for each new iteration of the loop. If Ooptj is negative, the value Oj is incremented and if Ootpj is positive, the value Oj is decremented. The rate distortion cost for Oj is calculated in step 903 according to the following formula:

Where λ is the Lagrange parameter and R(Oj) is a function, which provides the number of bits required for the codeword of Oj.

The formula'SumNbPixj×Oj×Oj-Sumj×Oj×2' gives an improvement based on the distortion representation provided by using the offset Oj. If J(Oj) is smaller than Jj, then in step 904, Jj=J(Oj) and ORDj is equal to Oj. If in step 905, Oj is equal to 0, the loop ends and the best ORDj for class j is selected.

The algorithms of Figures 18 and 19 provide the best ORDj for each class j. This algorithm is repeated for each of the four directions of FIG. 7A. Then, the direction system that provides the best rate distortion cost (sum of Jj for each direction) is selected as the direction to be used for the current CTU.

The algorithm for the selected offset value (Figures 18 and 19) for the encoder side of the edge offset tool can be easily applied to the band offset filter to select the best position (SAO_band_position), where j is the interval [0,32 ] Instead of the interval [1,4] in Figure 13. It involves changing the value 4 in modules 801, 810, 811 to 32. More specifically, for the 32 categories in FIG. 8, the parameter Sumj (j=[0,32]) is calculated. This is equivalent to calculating each range j, the difference between the current pixel value (Pi) and its original value (Porgi), and each pixel of the image belonging to a single range j. Then, using the same procedure as described in FIG. 14, the optimal offset according to the rate-distortion ORDj for 32 classes is calculated.

The next step involves finding the best position for the SAO band position of Figure 8. This is determined by the coding procedure described in Figure 15. The RD cost Jj for each range has been calculated with the coding procedure of FIG. 14 using the optimal offset ORDj based on the rate distortion. In FIG. 15, in the initial step 1001, the rate distortion value J is initialized to the maximum possible value. Then, a loop at 28 positions j of four consecutive classes is executed in step 1002. Furthermore, in step 1003, the variable Jj corresponding to the RD cost of the (4 consecutive classes) band is initialized to zero. Then, four loops of successive offset j are executed in step 1004. In step 1005, Jj is incremented by the RD cost of four types of Jj (j=i to i+4).

If this cost Ji is lower than the optimal RD cost J, then J is set to Ji, and in step 1007, sao_band_postion=i, and the next step is step 1008.

Otherwise, the next step is step 1008.

In the document JVET-J0024, it is proposed to modify the SAO edge classification. The following table illustrates the newly proposed SAO edge classification.

In this table, c is the value of sample c in FIGS. 7A and 7B, a is the value of the “left” sample corresponding to Cn1 in FIGS. 7A and 7B, and b is the value corresponding to FIGS. 7A and 7B The value of the "right" sample of Cn1. The symbol && is the operator AND. The symbol ‖ is the operator or (OR).

If this proposed modification is compared with the SAO edge offset classification of HEVC provided in the following table, the modification is only a modification of the sign function calculation.

其中函數sign(.)的定義由以下關係式所給定： sign(val)=-1，如果“val”＜-3 sign(val)=0，如果-3=“val”＜=3 sign(val)=+1，如果“val”＞3。Indeed, in order to obtain the edge index value given by the formula edge index Edgelndex =sign(C-Cn2)-sign(Cn1-C)+2, the sign function is modified by the following formula:

The definition of the function sign(.) is given by the following relationship: sign(val)=-1, if "val"<-3 sign(val)=0, if -3="val"<=3 sign( val)=+1, if "val">3.

And, where "test?val1:val2" is a ternary operator. The variable "test" is the status. If "test" is true, then "val1" is the returned value. If "test" is false, then "val2" is the returned value.

Unfortunately, this method does not improve the coding efficiency of VTM software for all coding architectures. For random access and internal, the gain obtained is less than 0.1%. For low-latency B frame, this method provides the average loss of test status with VTM software. The only architecture that may be improved is the low-latency P architecture.

First group of embodiments (first aspect)

According to the first aspect of the present invention, a method for obtaining an edge offset is provided to perform sample adaptive offset (SAO) filtering, wherein the "symbol" parameter N is sent to determine the index of the offset .

In other words, this first aspect relates to a method of performing sample adaptive offset (SAO) filtering on an image including a majority of image parts. The method includes: Get the edge offset to perform SAO filtering; Using a predetermined symbol method to determine the index of the edge offset, applying an edge offset classification to at least one pixel of the image; The symbol method predetermined to determine the index of the edge offset; The predetermined symbol method is sent in the bit stream.

According to this first aspect, the encoder selects a symbol method among the most possible symbol methods. The best symbol method is signaled in the bitstream, so that the decoder applies the relevant edge offset classification in this symbol selection method.

There are several possibilities to define a symbolic method. One of the possibilities may define the sign function "sign", whose consideration corresponds to the number N of this limit value. This symbolic function can be given, for example, by the following formula: SignN(val,N)=(val＜-N?-1: (val>N?1:0)) The definition of the function signN(.,.) can be given by the following relationship: signN(val,N)=-1, if "val"<-N, signN(val,N)=0, if -N <= "val" <=N, signN(val,N)=+1, if "val">N.

At this time, the edge index of the SAO edge offset classification can be obtained according to the following formula: Edge index=signN(C-Cn2,N)-signN(C-Cn2,N)+2

And, the SAO edge classification can become:

The advantage of this embodiment may be that the coding efficiency is increased compared to traditional HEVC edge offset classification. Therefore, this first embodiment is more flexible in the prior art. Therefore, the increase in coding efficiency provided by the first embodiment is due to additional flexibility, which is suitable for quantization errors and video content.

The function signN is just an example of an implementation, and another possibility may hold the same sign function as defined by HEVC and divide the input value "val" by the number N. In this example, the edge index of the SAO edge offset classification can be obtained according to the following formula: Edge index=sign((C-Cn2)/N)-sign((C-Cn2)/N)+2.

According to other embodiments, those skilled in the art can change this division by multiplication and the following displacements: Edge index=sign(((C-Cn2)*R)>>L)-sign(((C-Cn2)*R)>>L)+2 Where R and L are derived from N values.

16 is a flowchart illustrating how to perform SAO filtering on the image portion according to the first embodiment of the first aspect of the present invention.

For example, the procedure 1600 may be executed during step 67 in FIG. 6. It can also be executed in the decoder's decoding loop (step 415 in Fig. 4) to generate a reference frame for motion evaluation and subsequent frame compensation. This procedure is about SAO filtering for a color component (hence, the superscript "_X" in the syntax element can be omitted as follows).

In this first embodiment, the predetermined symbology is transmitted in the bit stream at the CTU level. Therefore, the transmission system is executed at the CTU level, that is, the image part to which the program 1600 is applied is the CTU.

Compared with the procedure shown in FIG. 11, this procedure includes two additional modules 1617 and 1618 as described below.

The initial step 1601 includes determining the SAO filtering parameters according to the procedures described in FIGS. 9 and 10. The SAO filtering parameters are determined by the encoder and the encoded SAO parameters are included in the bitstream. Therefore, on the decoder side of step 1601, the decoder reads and decodes the parameters from the bitstream. Step 1601 takes sao_type_idx and if it is equal to 1, also takes sao_band_position 1602 and if it equals 2, it also takes sao_eo_class_luma or sao_eo_class_chroma (according to the color component processing). If the element sao_type_idx is equal to 0, no SAO filtering is applied.

At step 1601, the decoder also checks whether the SAO type is edge offset (EO). According to the first embodiment, a flag is inserted into the bit stream to offset each SAO edge at the CTU level. For this purpose, it defines the index eo_sign_method representing the value N, which will be used as the second parameter of the signN function of the edge classification of the current CTB sample. Furthermore, eo_sign_method was linked to eo_sign_method_Luma in 1618. between eo_sign_method_Chroma. As for other edge offset parameters, eo_sign_method_Luma can be independent of eo_sign_method_Chroma. Step 1601 also obtains an offset table 1603 for four offsets.

The variable i of each pixel Pi used to continuously consider the current block or frame area (CTU) is set to 0 in step 1604. In addition, "frame area" and "video area" are used interchangeably in this manual. In this example, the frame area is CTU. In step 1606, the pixel Pi is extracted from the frame area 1605 containing N pixels. This pixel Pi is classified in step 1607 according to the edge offset classification described with reference to FIG. 27A or 27B or the band offset classification described with reference to FIG. 8. It is decided that the module 1608 tests whether the Pi is in the class to be filtered using traditional SAO filtering.

If Pi is in a filtering class, the number of relevant classes j is specified and in step 1610, the relevant offset value Offsetj is extracted from the offset table 1603. In the case of conventional SAO filtering, this Offsetj is then added to the pixel value Pi in step 1611 to produce a filtered pixel value Pi' 1612. In step 1613, this filtered pixel Pi' is inserted into the filtered picture frame area 1616.

If Pi is not in the category for SAO filtering, Pi (1609) will be inserted into the filtering frame area 1616 in step 1613 without filtering.

After step 1613, the variable i is incremented in step 1614 to filter subsequent pixels of the current frame area 1605 (if any-test 1615). At step 1615, if all pixels have been processed (i>=N), the filtered frame area 1616 is reconstructed and can be added to the SAO reconstruction frame (see frame 68 in FIG. 6 or frame 416 in FIG. 4) .

FIG. 17 illustrates a procedure of parsing the SAO parameters in the bit stream 1701 on the decoder side according to the first embodiment shown in FIG. 16.

Compared with the procedure shown in FIG. 10, this procedure includes the following new module 1711.

In the initial step 1702, the "sao_type_idx_X" syntax element is read and decoded. The codeword representing this syntax element can use a fixed-length code or can use any arithmetic coding method. This syntax element sao_type_idx_X completes the determination of the color component Y or the two-chrominance component U&V of the type of SAO applied to the frame area. The technology of the present invention is not limited to YUV color components but can also be used for RGB color components or any other color components.

In the same step 1702, a test is performed to determine whether "sao_type_idx_X" is strictly positive. If "sao_type_idx_X" is equal to "0", it means that for Y, if X is set equal to Y, there is no SAO in this frame area (CTU) and that for U and V if X is set equal to U and V, there is no SAO in this frame area . The decision of the SAO parameters is completed and the procedure proceeds to step 1708. Otherwise, if "sao_type_idx" is strictly positive, it means that there is a SAO parameter for this CTU in the bitstream.

Then, the procedure proceeds to step 1706, where the loop performs four iterations. Four iterations are performed in step 1707, where the absolute value of offset j is read and decoded by the bit stream. These four offsets correspond to the four absolute values of the four edge index offsets (O1, O2, O3, O4) of the SAO edge offset (Figure 27A or 27B) or correspond to the four related SAO band offsets. Four absolute values of the range of offsets (see Figure 8).

After step 1707, the process proceeds to step 1703, where a test is performed to determine whether the SAO type should be equal to the offset type (sao_type_idx_X==1).

If the result is positive, the symbol used for the offset with offset mode is decoded in steps 1709 and 1710, except for each offset with a value of zero, before the subsequent step 1704 is performed, in the bit stream The position “sao_band_position_X” of the SAO band is read and decoded as shown in FIG. 8.

If the result in step 1703 is negative ("Sao_type_idx_X" is set equal to 2), indicating that the edge offset type is used. According to this embodiment, the decoder in this example reads eo_sign_method_X from the bitstream 1701 in step 1711. If X is equal to Y, the read syntax element is "eo_sign_method_luma" and if X is set equal to U and V, the read syntax element is "eo_sign_method_chroma".

It can be noted that in this example, because some other parameters are shared between the two chroma components, eo_sign_method_X is used for the two chroma components (U and V). However, if the input video signal type is RGB, individual eo_sign_method syntax elements may also be specially considered.

The eo_sign_method syntax element may be encoded with a fixed-length code representing the number of possible values of parameter N. It has been decided that the four possible values for N can provide an interesting compromise between coding complexity and coding efficiency. Therefore, this syntax element can be encoded in 2 bits.

Furthermore, the edge offset class (corresponding to directions 0, 45, 90, and 135 degrees) is extracted from the bit stream 1701 in step 1705. If X is equal to Y, the syntax element is read as "sao_eo_class_luma" and if X is set equal to U and V, the syntax element is read as "sao_eo_class_chroma".

When the four offsets are decoded, the reading of the SAO parameters is completed and the process proceeds to step 1708.

The advantage of the CTU level signal edge symbol method is based on the effective adaptation of the frame/slice content and quantization error, which results in an improvement in coding efficiency compared to previous techniques.

According to a possible implementation method, the predetermined sign method is CABAC coded, that is, the flag eo_sign_method_X may be coded in traditional CABAC arithmetic. The advantage of this implementation method may be that although the frame/slice content and quantization error in the current slice/picture frame are relatively homogeneous, the coding efficiency can still be increased. As a result, an edge offset sign method can be selected for many CTUs.

FIG. 18 illustrates a procedure of parsing the SAO parameters in the bit stream 1801 on the decoder side according to the second embodiment.

According to the second embodiment, the predetermined symbol method is predicted. More specifically, the edge offset symbol method syntax element eo_sign_method_X can be predicted by another symbol method syntax element.

Compared with the procedure shown in FIG. 17, this procedure includes two additional modules 1812 and 1813 as described below.

In the initial step 1802, the "sao_type_idx_X" syntax element is read and decoded. The codeword representing this syntax element may use a fixed-length code or may use any method of arithmetic coding. The syntax element sao_type_idx_X completes the SAO type determination applied to the frame area of the color component Y or the two-chrominance component U&V to be processed. The technique of the present invention is not limited to use with YUV color components, but can also be used with RGB color components or any other color components.

In the same step 1802, a test is performed to determine whether "sao_type_idx_X" is strictly positive. If "sao_type_idx_X" is equal to "0", it means that for Y if X is set equal to Y, there is no SAO for this frame area (CTU), and for U and V, if X is set equal to U and V, there is no SAO for This picture frame area. The decision of the SAO parameter is completed and the procedure proceeds to step 1808. Otherwise, if "sao_type_idx" is strictly positive, it means that the SAO parameter exists in this CTU of the bitstream.

Then, the procedure proceeds to step 1806, where the loop performs four iterations. Four iterations are performed in step 1807, where the absolute value of offset j is read and decoded by the bit stream. These four offsets correspond to the four absolute values of the four edge index offsets (O1, O2, O3, O4) of the SAO edge offset (see Figure 27A or 27B) or the four corresponding to the SAO band offset The four absolute values of the range-dependent offset (see Figure 8).

After step 1807, the process proceeds to step 1803, where a test is performed to determine that the type of SAO corresponds to the band offset type (sao_type_idx_X==1).

If the result is positive, the symbol for the offset with offset mode is decoded in steps 1809 and 1810, except for each offset with a value of zero, before the subsequent step 1804 is executed to read the bit The position "sao_band_position_X" of the stream and decoded SAO band is shown in Fig. 8.

If the result is negative in step 1803 ("Sao_type_idx_X" is set equal to 2), indicating that the edge offset type is used. The decoder in this example reads eo_sign_method_X from the bitstream 1801 in step 1811. If X is equal to Y, the read syntax element is "eo_sign_method_luma" and if X is set equal to U and V, the read syntax element is "eo_sign_method_chroma".

Then, according to this second embodiment, the predictor symbol method 1812 for the color component X is added to the decoding syntax element eo_sign_method_X in step 1813 to obtain the edge offset symbol method for the current CTB.

Therefore, on the encoder side, Predictor_sign_method_X has been subtracted from eo_sign_method_X.

In this second embodiment, the syntax element eo_sign_method_X is coded with variable-length codes to obtain the advantages expressed in coding efficiency. The variable-length code may be a unary code or a Golomb code or another method. This Predictor_sign_method_X can be used as the context value for CABAC encoding.

According to an alternative embodiment, the flag may indicate whether the current eo_sign_method_X is strictly equal to Predictor_sign_method_X or not. If not, eo_sign_method_X can be decoded. otherwise, eo_sign_method_X is equal to Predictor_sign_method_X. Therefore, eo_sign_method_X can be a fixed-length code.

In a sub-embodiment, if the left CTU exists (for example, if the left CTU is used as the SAO parameter and if the SAO type is an edge offset type), then eo_sign_method_X may be predicted by the eo_sign_method_X of the left CTU.

In a sub-embodiment, if the upper CTU exists, eo_sign_method_X may be predicted by eo_sign_method_X of the upper CTU.

These two sub-embodiments have the same advantages as the SAO merge derivation, and their current SAO parameters for the left and upper CTUs.

In a sub-embodiment, if any, eo_sign_method_X can be the latest decoded previously Predicted by eo_sign_method_X.

In a sub-embodiment, if any, eo_sign_metod_X can be a few previously decoded The average predicted by eo_sign_method_X.

After step 1813, the edge offset classes (corresponding to directions 0, 45, 90 and 135 degrees) are extracted from the bit stream 1801 in step 1805. If X is equal to Y, the read syntax element is "sao_eo_class_luma" and if X is set equal to U and V, the read syntax element is "sao_eo_class_chroma".

When the four offsets have been decoded, the reading of the SAO parameters is completed and the process proceeds to step 1808.

The advantage of the second embodiment and all its sub-embodiments is that the coding efficiency is increased due to the use of the spatial correlation of the signal content and the similar quantization error between the CTUs of the current slice/picture frame. This is because when the signal of the current slice is similar by CTU, and when the quantization error is also similar, the best choice eo_sign_method_X on the encoder side is often similar.

FIG. 19 illustrates a procedure of parsing the SAO parameters in the bit stream 1901 on the decoder side according to the third embodiment.

According to this third embodiment, the edge offset symbol method is transmitted at a higher level than the CTU level. In other words, the predetermined symbology is signaled in the bitstream with frame, slice or sequence levels.

More specifically, it sends messages in slices/picture frames and utilizes the possible correlation between the edge-offset symbol methods used in slices/picture frames.

At step 1902, slice_sao_luma_flag is extracted from the bitstream 1901. If the value of the flag is set to true (=1), in step 1903, eo_sign_method_luma is extracted from the bitstream 1901 and will be used for the edge offset classification notation of all luminance CTBs of the current slice/picture frame. Then, the program will proceed to the next slice header syntax element (step 1906).

If the value of the flag slice_sao_luma_flag is set to false, then in step 1904, slice_sao_chroma_flag will be extracted from the bitstream 1901. If the value of this flag is set to true (=1), eo_sign_method_chroma is extracted from the bitstream 1901 at step 1905 and will be used for the edge offset classification notation for all chroma CTBs of the current slice/picture frame. Then, the program proceeds to the next slice header syntax element (step 1906). If the value of the flag is set to false (=0), the program will directly proceed to the next slice header syntax element (step 1906).

The advantage of this embodiment may be that when the frame/slice content and quantization error are relatively homogeneous within the current slice/picture frame, the coding efficiency is improved. The main advantage of sending messages in a slice file is that the number of possible values for eo_sign_method_X may be higher than the CTU level signaling method described with reference to FIGS. 17 and 18. This is because the code rate of the slice header may have a lower impact than the syntax elements of the CTU level. Therefore, eo_sign_method_X can be encoded in 3 bits (8 possible values) or 4 bits (16 possible values).

According to an alternative embodiment, eo_sign_method_X can be sent at a sequence level. In this case, the quantization error and signal content should be similar to frame-by-frame and CTU-by-frame.

According to an embodiment, the predetermined symbol method is selected on the encoder side to minimize the rate-distortion cost. Therefore, for each The different possible values of eo_sign_method_X are evaluated using traditional RD criteria and the value that minimizes the RD (rate-distortion) cost is selected and inserted into the bitstream.

The advantage of this embodiment may be that it should reach the maximum coding efficiency of the slice/picture frame level signaling.

FIG. 20 illustrates an example of selection of eo_sign_method_X by the encoder. It represents the picture frame hierarchy based on the depth at slice level (encoder parameters only) and quantization parameters. As shown in this figure, as the quantization parameter increases, the eo_sign_method_X value increases.

According to this embodiment, the predetermined symbol method is selected on the encoder side according to the parameters of the current slice or picture frame. In other words, the encoder can choose according to one or more parameters of the current slice/picture frame eo_sign_method_X. These parameters can be depth or quantization parameters in the picture frame hierarchy as shown.

According to an alternative embodiment, any eo_sign_method_X, for example, is transmitted immediately at the beginning of the sequence. In this case, eo_sign_method_X is selected or pre-selected in a sequence level manner to define a codec to be optimized for the individual components or luminance components and dichromatic components of all sequences.

FIG. 21 is a flowchart illustrating how SAO filtering is performed on the image portion according to the fourth embodiment of the first aspect of the present invention.

For example, this procedure 2100 is executed during step 67 of FIG. 6. It is also executed in the decoding loop of the encoder (step 415 of FIG. 4) to generate a reference picture frame for motion evaluation and compensation of subsequent picture frames. This procedure is related to SAO filtering for a color component (this subscript "_X" in the syntax element has been omitted as follows).

In this fourth embodiment, the obtained eo_sign_method_X does not directly correspond to the value N used as a parameter in the signN function. The table system is used to convert the index value eo_sign_method_X is converted to this value N. For example, according to a majority limit value N that is not continuous and/or may not contain 0 as a possible value, on the encoder side, a predetermined sign method is selected.

Compared with the procedure shown in FIG. 16, the procedure includes two additional modules 2119 and 2120 as described below.

The initial step 2101 includes determining the SAO filter parameters according to the procedures depicted in FIGS. 9 and 10. The SAO filtering parameters are determined by the encoder and the encoded SAO parameters are included in the bitstream. Therefore, on the decoder side of step 2101, the decoder reads and decodes the parameters from the bitstream. Step 2101 obtain sao_type_idx and if it is equal to 1, then also obtain sao_band_position 2102, if it is equal to 2, it also obtains sao_eo_class_luma (depending on the color component being processed) or sao_eo_class_chroma. If the element sao_type_idx is equal to 0, no SAO filtering is applied.

In step 2101, the decoder also checks whether the SAO type is edge offset (EO). For each SAO edge offset at the CTU level, a flag is inserted inside the bit stream. For this purpose, it defines an index eo_sign_method representing the value N, which will be used as the second parameter of the signN function of the edge classification of the current CTB sampling. Furthermore, eo_sign_method is specified between eo_sign_method_Luma and eo_sign_method_Chroma (step 2118). As for the other edge offset parameters, eo_sign_method_Luma may not be related to eo_sign_method_Chroma. Step 2101 also obtains an offset table 2103 of 4 offsets.

Furthermore, according to this embodiment, the new syntax element M is based on the syntax element eo_sign_method_luma or o_edge_sign_method_chroma and the conversion table "Set" are defined in step 2119. The table set is used to convert the index value eo_sign_method_X to the value M. An example of this conversion table is provided below. This value M will be used in the function "signN(x,M)" 2120 for the edge offset classification of the current CTB samples.

The variable i used to continuously consider each pixel Pi of the current block or frame area (CTU) is set to 0 in step 2104. In addition, "frame area" and "video area" are used interchangeably in this manual. The frame area in this example is CTU. In step 2106, the pixel Pi is extracted from the frame area 2105, and the frame area includes N pixels. In step 2107, the pixels Pi are classified according to the edge offset classification described with reference to FIGS. 27A and 27B or the band offset classification described with reference to FIG. It is decided that the module 2108 tests whether Pi is in a category to be filtered using traditional SAO filtering.

If Pi is in the filtering class, the relevant class number j is specified and the relevant offset value Offsetj is extracted from the offset table 2103 at step 2110. In conventional SAO filtering, this Offsetj is then added to the pixel value Pi in step 2111 to produce a filtered pixel value Pi' 2112. This filtered pixel Pi' is inserted into the filtered frame area 2116 in step 2113.

If Pi is not in a category that is subject to SAO filtering, Pi (2109) is inserted into the filtered frame area 2116 in step 2113 without filtering.

After step 2113, the variable i is incremented in step 2114 to filter the subsequent pixels of the current frame area 2105 (if any-test 2115). After all pixels are processed in step 2115 (i>=N), the filtered frame area 2116 is reconstructed and can be added to the SAO reconstruction frame (see frame 68 in FIG. 6 or 416 in FIG. 4).

For ease, the following example contains the four values in the set of possible N. The group may have N as the value of the table group. However, the present invention is not limited to this.

For example, the value N=0 can be excluded, and the set of N values used can be {1,2,3,4}. This can provide better RD compromise.

Alternatively, the set of possible values of N can be aligned with the shift value of the value Val. For example, the group may be {0, 2, 4, 8}. It corresponds to the right shift M (using M={0,1,2,3}) using the classification symbol method and the input value val.

For example, the edge index can be obtained by the following formula:

It should be noted that this formula does not always provide the exact same result, as when N between the groups {0, 2, 4, 8} is used with the following signN function:

This is due to the right shift operator on negative values. But the coding efficiency of the two methods is similar.

其中eo_sign_method_X可以採4個值0，1，2，3。此組使得其更容易由eo_sign_method_X推導出值N。In an embodiment, the set of possible N values may be {1, 2, 4, 8}. In this case, the edge index can be obtained by the following formula:

Among them, eo_sign_method_X can take 4 values 0, 1, 2, 3. This group makes it easier to derive the value N from eo_sign_method_X.

In other embodiments, the group N may be adapted to match the statistics of the signal content as {0, 2, 4, 6} or another example as {0, 1, 3, 6}.

In these examples, the set of possible values N is established so that the maximum value is not very large and the difference between the first value of the set and the last value of the set is relatively small.

When the transmission is at the CTU level or lower, all these embodiments may be interesting because continuous values do not correspond to statistics of natural content and related quantization errors. One advantage is the improvement of coding efficiency.

According to another embodiment, the predetermined symbol method is selected on the encoder side according to a majority limit value N that depends on the slice type, that is, the set of N values may depend on the slice type. Because the content of the quantization error depends on the slice type, it may be interesting even if the subjective measurements give the same quality measurement. In this embodiment, the value contained in the group between P slices is higher than the value used for the group between B slices.

In an alternative embodiment, at least one of the set of possible N values used between P slices is higher than the corresponding value within the set of possible N values used between B slices. The N value may only be related to luminance (Luma).

其中Slice_type為內、P間或B間。This alternative example can be implemented by modifying the procedure shown in FIG. 21, for example, by using this formula in module 2119:

Where Slice_type is inner, P or B.

For example, the table group can be as follows:

According to another embodiment, the predetermined symbol method is selected on the encoder side according to a majority limit value N. The limit value depends on the color component, that is, the set of N values depends on the color component. In this embodiment, the set of possible values N for the luminance component may be different from the set of possible values N for the two-chroma component. The advantage of this embodiment may be the improvement in coding efficiency due to the adaptation group N value for each color component type.

In an additional embodiment, the set of N values for luma components contains higher values than the set of N values for chroma components. The advantage of this embodiment may be the improvement in coding efficiency due to the adaptation group N value for each color component type. This is because for the chrominance component, the difference between adjacent samples is lower than the luminance component, so it is preferable to lower the value N for chrominance rather than the luminance component.

In an additional or alternative embodiment, the number of possible N values in the group for chroma is lower than the number of possible N values in the group for brightness. The advantage of this embodiment may be because there is a method of reducing the code rate of the signal transmission with the edge offset symbol method to a better number, so the coding efficiency is increased.

In one embodiment, when the input signal is YUV, two sets of N values are used, and when the input signal is RGB, one set of N values is used.

其中X等於亮度或色度，及Slice_type為內、P間或B間。These embodiments can be implemented by modifying the procedure in FIG. 21, for example, by using this formula in module 2119:

Where X is equal to brightness or chroma, and Slice_type is inner, P or B.

For example, the table group can be as follows:

FIG. 22 is a flowchart illustrating how the SAO filtering system is performed on the image portion according to the fifth embodiment of the first aspect of the present invention.

According to this fifth embodiment, the predetermined sign method is selected on the encoder side based on the majority limit value N that only depends on the luminance component, that is, the N value is only related to the luminance component. In practice, the set of possible N values for chroma components is therefore set to zero. The advantage of this embodiment can be improved coding efficiency by removing eo_sign_method_chroma. This embodiment can be implemented by removing eo_sign_method_chroma from the aforementioned diagram.

Compared with the procedure shown in FIG. 17, this procedure includes an additional module 2212, which will be described as follows.

In the initial step 2202, the "sao_type_idx_X" syntax element is read and decoded. The codeword representing this syntax element may use a fixed-length code or may use any arithmetic coding method. The syntax element sao_type_idx_X causes the determination of the type of SAO applied to the picture frame area to be processed with the color component Y or with both chrominance components U&V. The technique of the present invention is not limited to use with YUV color components and can also be used with RGB color components or any other color components.

In the same step 2202, a test is performed to determine whether "sao_type_idx_X" is strictly positive. If "sao_type_idx_X" is equal to "0", it means that if X is set equal to Y, there is no SAO for this frame area (CTU) for Y, and, if X is set equal to U and V, then for U and V, there is no SAO is used for this picture frame area. The decision of the SAO parameter is completed and the process proceeds to step 2208. Otherwise, if "sao_type_idx" is strictly positive, it means that there is a SAO parameter for this CTU in the bitstream.

Then, the procedure proceeds to step 2206, where the loop performs four iterations. Four iterations are performed in step 2207, where the absolute value of the offset j is read and decoded by the bit stream. These four offsets correspond to the four absolute values of the offsets (O1, O2, O3, O4) of the four edge indexes of the SAO edge offset (see FIG. 27A or 27B) or four with the SAO band offset Four absolute values of offset related to range (see Figure 8).

After step 2207, the process proceeds to step 2203, where a test is performed to determine whether the type of SAO corresponds to the offset type (sao_type_idx_X==1).

If the result is positive, the symbol for the offset with offset mode is decoded in steps 2209 and 2210, except for each offset bit with a value of zero, to read the in-position before the following step 2204 is executed In the meta-stream, the position "sao_band_position_X" of the SAO band shown in FIG. 8 is decoded.

If the result in step 2203 is negative ("Sao_type_idx_X is set to 2), it means that the edge offset type is used. According to this embodiment, in step 2212, the decoder checks whether the component is considered to be a luminance component, and if so, reads the bit from bit in step 2211 Eo_sign_method_Luma of stream 2201. Otherwise, the procedure directly proceeds to step 2205.

Furthermore, the edge offset class (corresponding to directions 0, 45, 90, and 135 degrees) is extracted from the bit stream 2201 in step 2205. If X is equal to Y, the syntax element is read as "sao_eo_class_luma" and if X is set equal to U and V, the syntax element is read as "sao_eo_class_chroma".

When the four offsets are decoded, the reading of the SAO parameters is completed and the process proceeds to step 2208.

FIG. 23 illustrates a procedure of parsing the SAO parameters in the bit stream 2301 on the decoder side according to the sixth embodiment.

According to this embodiment, the predetermined symbol method is selected on the encoder side based on the majority limit value N transmitted in the slice header, that is, only one set of N values is transmitted in the slice header for luminance components and only One set is transmitted for chroma components. An advantage of this embodiment can be to increase coding efficiency by setting a useful set for each slice on the encoder side.

In step 2302, slice_sao_luma_flag is extracted from the bit stream 2301. If the value of the flag is set to true (=1), then Max_eo_sign_method_luma is extracted from the bit stream 2301 in step 2303. This value represents the number of components in the group. Then, each element value (2305) of the group is extracted (2307) from the bit stream (2301) and will be used for the edge offset classification notation of all brightness CTBs of the current slice/picture frame. The process then proceeds to the next slice header syntax element (step 2306).

If the value of the flag slice_sao_luma_flag is set to false, slice_sao_chroma_flag is extracted by the bit stream 2301 at step 2304. If the value of the flag is set to true (=1), then Max_eo_sign_method_chroma is extracted by the bit stream 2301 at step 2308. This value represents the number of components in the group. Then, each element value (2309) in the group is extracted (2310) from the bit stream (2301) and will be used for the edge offset classification notation of all brightness CTBs of the current slice/picture frame. The process then proceeds to the next slice header syntax element (step 2306). If the value of the flag is set to false (=0), the program directly proceeds to the next slice header syntax element (step 2306).

FIG. 24 is a flowchart illustrating step 240 of a statistical calculation example that can be applied to an edge offset filter on the encoder side in the case of SAO filtering according to an embodiment of the present invention.

In this embodiment, there is a competition between different N values of the encoder. In these embodiments, the RD cost of each possible value of eo_sign_method is calculated to select the best on the encoder side eo_sign_method.

This figure illustrates the setting of the variable CTUStats that contains all the information needed to derive the best bit rate distortion offset for each class. Furthermore, it illustrates the selection of the best SAO parameter set for the current CTU.

For each color component Y, U, B (or RGB) (2411), each SAO type is evaluated. For each SAO type (2412) and each eo_sign_method_X (2416), in the initial step 2401, the variables Sum _j and SumNbPix _j are set to zero. The current picture frame area 2403 includes N pixels. j is the current range number, used to determine the four edge indexes for the edge offset type shown in FIG. 7B, or the 32 ranges for the pixel values with offset type shown in FIG. 8) . Sum _j is the sum of the differences between pixels in the range j and their original pixels. SumNbPix _j is the number of pixels in the frame area, and its pixel value belongs to the range j.

In step 2402, the variable i used to continuously consider each pixel Pi of the current frame area is set to zero. Then, the first pixel P _{i of the} picture frame area 2403 is extracted in step 2404. In step 2405, the type of the current pixel is determined by checking the state defined in FIGS. 27A and 27B. Then, in step 2405, a test is performed. During step 2405, a check is performed to determine whether the pixel P _i value corresponds to the value of class "None of the above" of FIGS. 27A and 27B.

If the result is positive, the value "i" is incremented in step 2408 to consider the next pixel of the frame area 2403.

間之差被加至Sum_j 。在下一步驟2408中，變數i被增量，以考量畫框區域2403的下一像素。Otherwise, if in step 2406, the result is negative, then the next step 2407, where the relevant SumNbPix _j (i.e., the sum of the number of pixels for the class determined at step 2405) P _i is incremented and in its Original value

The difference between them is added to Sum _j . In the next step 2408, the variable i is incremented to consider the next pixel of the frame area 2403.

Then, test 2409 is executed to determine whether all pixels have been considered and classified. If the result is negative, the procedure returns to step 2404 above. Otherwise, the result is positive, the routine proceeds to step 2410, where a variable current CTUStats color component X, the SAO type SAO _- _type, eo_sign_method_X and existing class j is set equal to a first value for the first and the Sum _j SumNbPix _{j with} two values.

The program then loops (2417) for each eo_sign_method_X until the last one (2418) and for the SAO type (2413) until the last one (2414). When the last one is reached, the program processes the next component (2415).

As described with reference to FIG. 13, these variables can be used to calculate, for example, the optimal offset parameter Offset _j for each category _j . The offset Offset _j may be the average of the difference between the pixels of this type j and its original value. Therefore, Offset _j is given by the following formula:

Note that Offset _j is an integer value. Therefore, the code rate defined in this formula can be rounded down to the nearest value or use the ceiling or floor function.

Each offset Offset _j is the optimal offset Ooptj when represented by distortion.

Therefore, in FIG. 24, since the new loop system on the edge offset symbol method is added (2416, 2417, 2418) to obtain the statistical CTUStats (2410) for each edge offset symbol method, the table CTUStats It has additional dimensions to set the statistics for the edge offset sign method compared to FIG. 13.

To evaluate the combined RD cost of SAO parameters, the encoder uses the statistical set in the table CTUCStats. According to the following example for SAO left union and by considering the type used for Luma Left_Type_Y and the four related offsets O_Left_O, O_Left_1, O_Left_2, O_Left_3, the distortion can be obtained by the following formula.

The variable Shift is designed for distortion adjustment. Distortion should be negative because SAO is positive filtering.

The same calculation can be applied to the chroma component. The rate distortion cost λ is fixed for these three components. For SAO parameters that are merged with the left CTU, the code rate is only 1 flag, which is CABAC encoded.

FIG. 25 is a flowchart illustrating the steps of performing an encoder to determine SAO parameters for a group (frame or slice) of CTUs.

In this embodiment, the predetermined symbol method is selected on the encoder side according to the majority limit value N depending on the color component. Therefore, the statistics obtained by the procedure of FIG. 24 can be used to determine the best edge class and the best eo_sign_method for the luminance and the two chroma components, respectively.

The procedure starts with the current CTU (2501). First, the statistical systems used for all possible SAO types and classes are accumulated in the variable CTUStats (2502). The procedure of step 2502 is described with reference to FIG. 24.

Based on the value set in the variable CTUStats, the RD cost for SAO left merge is evaluated, whether the left CTU is used as the RD cost for SAO merge (2504) in the current slice (2503). Due to the statistics in CTUStats (2502), the new SAO parameters are evaluated for brightness (2505), dichromatic components (2509), and each eo_sign_method_chroma (2518).

For each SAO type (2506) and each eo_sign_method_luma (2517), the best RD offset for offset classification and other parameters are obtained (2507). Due to its individual SAO parameters and luminance sign method (2508), all RD costs are calculated. For the two chroma components in the same way, the optimal RD offset and parameter system are selected (2511). Due to its individual SAO parameters and chromatic sign method (2512), all RD costs are calculated. All this RD costs are then compared to select the best SAO parameter set (2515). These RD costs are also compared, with SAOs that can be used independently for the luminance and chrominance components (2513, 2514). The use of the new SAO parameter set (2515) is compared to the SAO parameter set "merged" or shared (2516) by the left and upper CTUs.

Second group of embodiments (second aspect)

According to the second aspect of the present invention, a method of performing sample adaptive offset (SAO) filtering to obtain an edge offset is provided, wherein when the "symbol" parameter N is strictly positive, the at least part of the offset symbol is Send the letter explicitly.

In other words, the second aspect provides a method of performing sample adaptive offset (SAO) filtering on an image including a majority of image parts. The method includes: Obtain most edge offsets to perform SAO filtering; The method further includes signaling at least one edge offset symbol of the majority edge offset in the bit stream.

According to the first embodiment of the second aspect, the sign of the edge offset is sent to all edge offsets of the majority of edge offsets, where the sign parameter N used in the sign function that determines the index of the edge offset Is positive.

According to other embodiments, when at least one of the following conditions is met, the edge shifted symbol is sent: -Classification uses the new edge symbol method for this edge offset; -The input value of the sign function is modified for edge offset classification; -The parameter N used in the sign function used to determine the index of the edge offset is strictly positive.

The advantage of these embodiments is that the coding efficiency of the implicit symbol signaling for the offset of the edge offset is increased when the value N of the symbol method is not zero. This is an unexpected result, because the offset implicit symbol signal of the edge offset classification is known to cause a reduction in coding efficiency with the traditional symbol method.

FIG. 26 is a flowchart illustrating a procedure of decoding SAO parameters according to the second embodiment of the second aspect.

According to this second embodiment, if all the sign parameters N used in the sign function that determines the index of the edge offset are higher than the predetermined threshold, the sign of the edge offset is sent for all edge offsets of the majority of edge offsets shift.

Therefore, the symbol is sent only when the symbol parameter N is high for all offsets, that is, when all the symbol parameters N used in the symbol function used to determine the index of these edge offsets are higher than a predetermined threshold.

Compared with the procedure shown in FIG. 10, this procedure performs an additional loop, which includes additional modules 2611, 2612, 2613, and 2614 as described below.

In the initial step 2602, the "sao_type_idx_X" syntax element is read and decoded. The code character representing this syntax element can use a fixed-length code or can use any arithmetic coding method. The syntax element sao_type_idx_X makes the type of SAO used in the picture frame area a decision to process with the color component Y or the two-chroma component U&V.

In addition, although YUV color components are used in HEVC (sometimes called Y, Cr, and Cb components), it will be understood that other color components of other video coding schemes may be used, for example, RGB color components. The technique of the present invention is not limited to use with YUV color components, and can also be used with RGB color components or any other color components.

In the same step 2602, a test is performed to determine whether "sao_type_idx_X" is strictly positive. If "sao_type_idx_X" is equal to "0", it means that if X is set equal to Y, then for Y, this frame area (CTU) has no SAO, and if X is set equal to U and V, it is used for U and V, and there is no SAO Used for this picture frame area. The determination of the SAO parameters is completed and the process proceeds to step 2608. Otherwise, if "sao_type_idx" is strictly positive, it means that there is a SAO parameter for this CTU in this bit stream.

Then, the process proceeds to step 2606, where the loop is executed four iterations. The four iterations are performed in step 2607, where the absolute value of offset j is read and decoded from the bit stream. These four offsets correspond to the four absolute values of the four edge index offsets (O1, O2, O3, O4) of the SAO edge offset (see FIGS. 27A or 27B) or correspond to the SAO band offset ( See Figure 8) for the absolute values of the four ranges of offset.

After step 2607, the process proceeds to step 2603, where a test is performed to determine whether the type of SAO corresponds to the band offset type (sao_type_ide_X==1).

If the result is positive, the symbol for the offset with offset mode is decoded in steps 2609 and 2610, except for each offset with a value of zero, before performing the following step 2604 to read the bit The position "sao_band_position_X" in the stream and the decoded SAO band is illustrated in FIG. 8.

If the result in step 2603 is negative ("Sao_tye_idx_X" is set equal to 2), which means that the edge offset type is used. If X equals Y, the read syntax element is "sao_eo_class_luma" and if X is set equal to U and V, the read syntax element is "Sao_eo_class_chroma". Therefore, the edge offset class (corresponding to directions 0, 45, 90, and 135 degrees) is extracted from the bit stream in step 2605.

When four offsets are extracted, the process proceeds to step 2611, where it tests whether the edge offset sign method eo_sign_method_X (2614), that is, the index system corresponding to the N value in the group table of the current color component is greater than the threshold Th (2611). If yes, then four symbols (2612) are extracted from the bit stream (2613). Then, the reading of the SAO parameters is completed and the procedure proceeds to step 2608.

The advantage of this embodiment may be that the encoding efficiency is increased compared to when the value N used in the symbol method is high, the implicit symbol used for edge offset classification is sent.

It can also be noted that in this embodiment, the edge offset symbol method may be that all CTUs used for the sequence are the same or selected slice by slice. When the edge offset sign method always has the same value, the module 2611 may not be used.

A description of the edge offset for these embodiments of the second aspect will now be provided with reference to FIGS. 27A and 27B.

The edge offset type involves determining the edge index for each pixel by comparing its pixel value with the values of two adjacent pixels. Furthermore, the two adjacent pixels depend on a parameter that indicates the direction of the two adjacent pixels relative to the current pixel. These directions are 0 degrees (horizontal direction), 45 degrees (diagonal direction), 90 degrees (vertical direction), and 135 degrees (second diagonal direction) as shown in FIG. 7A.

FIG. 27A shows a table of offset values to be applied to pixel values of specific pixels on the decoder side, when the value N is greater than the threshold Th.

When the value of C is lower than the two values of the adjacent pixels Cn1 and Cn2, the offset of the pixel value added to the pixel C is "O1" (O1 may be positive or negative or equal to 0). When the pixel value of C is lower than a pixel value of the adjacent pixel (Cn1 or Cn2) and C is equal to a value of the adjacent pixel, the offset added to the sample value of this pixel is "O2" (O2 can be positive or negative Or equal to 0).

When the pixel value of c is lower than one of the pixel values of the neighboring pixels (Cn1 or Cn2) and the pixel value of C is equal to a value of the neighboring pixels, the offset applied to this pixel sampling is "O3" (O3 can be Positive or negative or equal to 0). When the value of C is greater than the two values of Cn1 or Cn2, the offset applied to this pixel sample is "O4" (O4 can be positive or negative or equal to 0).

When none of the above conditions match the current sampling and its neighboring pixels, no offset value is added to the current pixel C, as described by the edge index value "2" in the table.

It is important to note that in the specific case of the edge offset type, the absolute value of each offset (O1, O2, O3, O4) is encoded in the bitstream. The symbol is also encoded.

In this FIG. 27A, the samples C, Cn1, and Cn2 correspond to those depicted in FIG. 7A.

FIG. 27B shows a table that gives the offset value applied to the pixel value of a specific pixel on the decoder side when the value N is lower than or equal to the threshold Th.

When the value of C is lower than the two values of the adjacent pixels Cn1 and Cn2, the offset of the pixel value added to the pixel C is "O1" (O1 may be positive or negative or equal to 0). When the pixel value of C is lower than a pixel value of the neighboring pixel (Cn1 or Cn2) and C is equal to a value of its neighboring pixel, the offset added to the sampled value of this pixel is "+O2" (this offset Is positive or equal to 0).

When the pixel value of c is lower than one of the pixel values of its neighboring pixels and the pixel value of C is equal to a value of its neighboring pixels, the offset applied to pixel sampling is "-O3" (this offset is negative Or equal to 0). When the value of C is greater than the two values of Cn1 or Cn2, the offset applied to this pixel sample is "O4" (O4 can be positive or negative or equal to 0).

When none of the above conditions match the current sampling and its neighboring pixels, the offset value is not added to the current pixel C, as plotted by the edge index value "2" in the table.

It is important to note that for the specific case of the edge offset type, the absolute value of each offset (O1, O2, O3, O4) is encoded in the bitstream. The symbol to be applied to each shift depends on the edge index to which the current pixel belongs (or the edge index in the HEVC specification). According to the table shown in FIG. 27B, for edge index 1 (O2), a positive offset is applied. For edge index 3 (O3), apply a negative offset to the current pixel. Corresponding to edge index 1 and edge index 4, symbols are encoded/decoded.

In this FIG. 27B, the samples C, Cn1, and Cn2 correspond to those depicted in FIG. 7A.

In additional embodiments, the symbols are CABAC encoded. In addition, no context is required. The advantage of this additional embodiment is that since the symbol is often equal to the implicit symbol, the coding efficiency is increased.

FIG. 28 is a flowchart illustrating a procedure for decoding SAO parameters according to the third embodiment of the second aspect.

According to this third embodiment, at least one edge offset for which the symbol is signaled belongs to the peak and valley structure. More specifically, as long as a predetermined symbol method is used to determine the edge offset index, at least one edge offset symbol is sent.

In this example, as long as the sign method is used to know the sign of the edge shift, the sign of the edge shift is explicit for peaks and valleys and implicit for the other two classes. This means that for the offsets O1 and O4 as shown in FIG. 27B, the symbols are transferred in the bit stream.

In the initial step 2802, the "sao_type_idx_X" syntax element is read and decoded. The code character representing this syntax element can use a fixed-length code or can use any arithmetic coding method. The syntax element sao_type_idx_X causes the SAO type determined to be applied to the frame area to be processed with the color component Y or the two-chrominance component U&V.

In addition, although YUV color components are used in HEVC (sometimes called Y, Cr, and Cb components), it should be understood that other video encoding schemes, other color components, such as RGB color components, can also be used. The technique of the present invention is not limited to use with YUV color components, but can also be used with RGB color components or any other color components.

In the same step 2802, a test is performed to determine whether "sao_type_idx_X" is strictly positive. If "sao_type_idx_X" equals "0", it means that if X is set equal to Y, there is no SAO for Y for this frame area (CTU), and if X is set equal to U and V, for U and V, this There is no SAO in the box area. The decision of the SAO parameter is completed and the procedure proceeds to step 2808. Otherwise, if "sao_type_idx" is strictly positive, it means that the SAO parameter exists in the CTU in the bitstream.

Then, the process proceeds to step 2806, where the loop is executed four iterations. The four iterations are performed in step 2807, where the absolute value of the offset j is read and decoded by the bit stream. These four offsets correspond to the four absolute values of the four edge index offsets (O1, O2, O3, O4) of the SAO edge offset (see Figure 27A or 27B) or to the SAO band offset (see Figure 8) Four absolute values of the offset associated with the four ranges.

After step 2807, the process proceeds to step 2803, where a test is performed to determine whether the type of SAO corresponds to the offset type (sao_type_idx_X==1).

If the result is positive, the symbol for the offset with offset mode is decoded in steps 2809 and 2810, except for each offset with a value of zero, to read the bit stream before the following step 2804 is executed The position of the decoded SAO band "sao_band_position_X" is shown in FIG. 8.

If the result in step 2803 is negative ("Sao_type_idx_X" is set equal to 2), indicating that the edge offset type is used. If X equals Y, read the syntax element as "sao_eo_class_luma" , And if X is set equal to U and V, the read syntax element is "Sao_eo_class_chroma". Therefore, the edge offset class (corresponding to directions 0, 45, 90, and 135 degrees) is extracted from the bit stream in step 2805.

For each offset (2812), determine whether the offset is a peak (j==1) or valley (j==4) offset, that is, O1 or O4 (2815). If so, the offset sign is extracted from the bit stream (2813). Otherwise, the symbol is implicit and set as shown in FIG. 27B. Then, the reading of the SAO parameters is completed and the procedure proceeds to step 2808.

The advantage of this embodiment may be that the coding efficiency of the implicit symbol signaling compared to the offset used for edge offset classification is increased, and the coding efficiency of the signaling transmission increased compared to the explicit symbol used for edge offset classification .

In another embodiment, the symbols are CABAC encoded. The advantage of this additional embodiment is increased coding efficiency because the symbol is often equal to the implicit symbol.

FIG. 29 is a flowchart illustrating a procedure of decoding SAO parameters according to the fourth embodiment of the second aspect.

According to this fourth embodiment, the symbols of edge offsets are only used for edge offsets, and the symbol parameter N used to determine the indexes of these edge offsets is higher than a predetermined threshold.

In other words, the sign of the edge offset is explicit for peaks and valleys (ie it is signaled in the bitstream) and implicit for the other two classes. Therefore, when N is greater than or equal to the threshold, the symbol is transmitted for offsets O1 and O4, as shown in FIG. 27B.

Compared to the procedure shown in FIG. 28, this procedure includes additional modules 2914 and 2911 as described below.

In the initial step 2902, the "sao_type_idx_X" syntax element is read and decoded. The codeword representing this syntax element can use a fixed-length code or any method that can use arithmetic coding. The syntax element sao_type_idx_X completes the application of the type of SAO applied to the frame area to be determined as the color component Y or the two-chrominance component U&V.

In addition, although YUV color components are used in HEVC (sometimes called Y, Cr, and Cb components), it should be understood that other video encoding schemes, other color components can also be used, such as RGB color components. The technique of the present invention is not limited to use with YUV color components, but can also be used with RGB color components or any other color components.

In the same step 2902, a test is performed to determine whether "sao_type_idx_X" is strictly positive. If "sao_type_idx_X" is equal to "0", it means that if X is set equal to Y, there is no SAO for this frame area (CTU) for Y, and if X is set equal to U and V, there is no frame area for U and V SAO. The decision of the SAO parameter is completed and the procedure proceeds to step 2908. Otherwise, if "sao_type_idx" is strictly positive, it means that there is a SAO parameter for this CTU in the bit stream.

Then, the procedure proceeds to step 2906, where the loop performs four iterations. Four iterations can be performed in step 2907, where the absolute value of the offset j is read and decoded by the bit stream. These four offsets correspond to the four absolute values of the four edge index offsets (O1, O2, O3, O4) of the SAO edge offset (see Figure 27A or 27B) or to the SAO band offset (see Figure 8) Four absolute values of offsets related to the four ranges.

In step 2907, the process proceeds to step 2903, where a test is performed to determine whether the type of SAO corresponds to the band offset type (sao_type_idx_X==1).

If the result is positive, the symbol for the offset with offset mode is decoded in steps 2909 and 2910, except for each offset with a value of zero, to read and decode the bit stream before the following step 2904 is executed The "sao_band_position_X" of the SAO band is shown in Figure 8.

If the result in step 2903 is negative ("Sao_type_idx_X" is set equal to 2), indicating that the edge offset type is used. If X equals Y, read the syntax element as "sao_eo_class_luma" , And if X is set equal to U and V, the read syntax element is "sao_eo_class_chroma". Therefore, the edge offset class (corresponding to directions 0, 45, 90, and 135 degrees) is extracted from the bit stream in step 2905.

When four offsets have been extracted, the process proceeds to step 2911, where it is tested whether the edge offset sign method eo_sign_method_X (2914), that is, the index system corresponding to the N value (in the conversion table "group") for the current color component is greater than the threshold Th (2911).

If so, for each of the four offsets (2912), decide whether the offset is a peak (j==1) or valley (j==4) offset, ie O1 or O4 (2815). If so, the offset sign is extracted from the bit stream (2913). Otherwise, the symbol is implicit and set as depicted in Figure 27B. Then, the reading of the SAO parameters is completed and the procedure proceeds to step 2908.

The advantage of this embodiment may be increased coding efficiency of implicit symbol signaling compared to edge offset classification and increased coding efficiency of explicit symbol signaling compared to edge offset classification. Compared with the third embodiment, the coding efficiency is also increased.

According to an embodiment, the symbols of the edge offset are only sent to the edge offset, where the symbol parameter N used to determine the index of these edge offsets is not zero.

According to these embodiments, the threshold value is equal to 0, so that when the sign method is the traditional sign method (N=0), the sign value is implicit for all offsets of the edge offset classification, and when the sign method has N> At 0, the sign used for the peak and valley offsets is explicit (that is, sent in the bit stream). This can provide very interesting coding efficiency improvements.

FIG. 30 is a flowchart illustrating a procedure of decoding SAO parameters according to the fifth embodiment of the second aspect.

According to the fifth embodiment, the symbol of the edge offset is only sent for the luminance component, that is, the offset symbol for the edge offset classification is explicit for the luminance component of FIG. 27A, and for the chrominance component of FIG. 7B is Implicitly.

This can provide coding efficiency improvements over explicit signaling for all color components.

In the initial step 3002, the "sao_type_idx_X" syntax element is read and decoded. The codeword representing this syntax element may use a fixed-length code or may use any arithmetic coding method. The syntax element sao_type_idx_X completes the decision that the SAO type applied to the frame area is processed as the color component Y or the two-chrominance component U&V.

In addition, although YUV color components are used in HEVC (sometimes called Y, Cr, or Cb components), it will be understood that other color coding schemes can use other color components, such as RGB color components. The technique of the present invention is not limited to use with YUV color components, but can also be used with RGB color components or any other color components.

In the same step 3002, a test is performed to determine whether "sao_type_idx_X" is strictly positive. If "sao_type_idx_X" is equal to "0", it means that if X is set equal to Y, then no SAO is used for this frame area (CTU) for Y, and if X is set equal to U&V, there is no SAO for U and V Used in this frame area. The decision of the SAO parameter is completed, and the process proceeds to step 3008. Otherwise, if "sao_type_idx "Is strictly positive, it means that the SAO parameter exists in this CTU of the bit stream.

Then, the procedure proceeds to step 3006, where the loop performs four iterations. The four iterations are performed in step 3007, where the offset j is read and decoded by the bit stream. These four offsets correspond to the four absolute values of the four edge index offsets (O1, O2, O3, O4) of the SAO edge offset (see Figures 27A and 27B) or to the SAO band offset (see Figure 8) Four absolute values of offsets related to the four ranges.

After step 3007, the process proceeds to step 3003, where a test is performed to determine whether the type of SAO corresponds to the offset type (sao_type_idx_X==1).

If the result is positive, the symbol for the offset with offset mode is decoded in steps 3009 and 3010, except for each offset with a value of zero, to read the bit stream before the following step 3004 is performed And decode the SAO band position "sao_band_position_X" as shown in FIG. 8.

If in step 3003, the result is negative ("Sao_type_idx_X" is set equal to 2), indicating that the edge offset type is used. If X equals Y, read the syntax element as "sao_eo_class_luma" , And if X is set equal to U and V, the read symbol element is "Sao_eo_class_chroma". Therefore, the edge offset class (corresponding to directions 0, 45, 90, and 135 degrees) is extracted from the bit stream in step 3005.

When the four offsets are extracted, the process proceeds to step 3016, which tests whether the current color component X corresponds to the luminance component. If yes, then four symbols (3012) are extracted from the bit stream (3013). Then, the reading of the SAO parameters is completed and the procedure proceeds to step 3008.

If the current color component is a chroma component, the process directly proceeds to step 3008.

The advantage of this embodiment may be improved coding efficiency compared to the transmission of symbols of all color components. This is because for chroma components, implicit signaling may be more efficient than explicit signaling of symbols.

In other embodiments, the symbol can signal all color components, which simplifies the hardware implementation.

FIG. 31 is a flowchart illustrating a procedure of decoding SAO parameters according to the sixth embodiment of the second aspect.

According to this sixth embodiment, when the SAO symbol method is used, that is, when the index corresponding to the N value in the "group" table is higher than the threshold Th, the symbol is explicitly signaled for brightness.

Compared with the procedure shown in FIG. 30, this procedure includes additional modules 3111 and 3114 as described below.

In the initial step 3102, the "sao_type_idx_X" syntax element is read and decoded. The code character representing this syntax element can use a fixed-length code or can use any arithmetic coding method. The syntax element sao_type_idx_X completes the decision that the type of SAO applied to the frame area is processed as the color component Y or both chrominance components U&V.

In addition, although YUV color components are used in HEVC (sometimes called Y, Cr, and Cb components), it should be understood that other color components of other video encoding schemes, such as RGB color components, can also be used. The technique of the present invention is not limited to use with YUV color components, but can also be used with RGB color components or any other color components.

In the same step 3102, a test is performed to determine whether "sao_type_idx_X" is strictly positive. If "sao_type_idx_X" is equal to "0", it means that if X is set equal to Y, then there is no SAO for Y used for this frame area (CTU), and if X is set equal to U and V, there is no SAO for U and V Used for this picture frame area. The decision of the SAO parameters is completed and the process proceeds to step 3108. Otherwise, if "sao_type_idx" is strictly positive, it means that there is a SAO parameter in the CTU in the bit stream.

Then, the process proceeds to step 3106, where the loop performs four iterations. Four iterations are performed in step 3107, where the absolute value of offset j is read and decoded by the bit stream. These four offsets correspond to the four absolute values of the offsets (O1, O2, O3, O4) of the four edge indexes of the SAO edge offset (see Figure 27A or 27B) or the four corresponding to the SAO band offset Four absolute values of offset related to range (see Figure 8).

After step 3107, the process proceeds to step 3103, where a test is performed to determine whether the type of SAO corresponds to the band offset type (sao_type_idx_X==1).

If the result is positive, the symbol for the offset with offset mode is decoded at steps 3109 and 3110, except for each offset with a value of zero, to read the bit stream before the following step 3104 is performed And the position "sao_band_position_X" of the decoded SAO band, as shown in FIG. 8.

If the result is negative in step 3103 ("Sao_type_idx_X" is set equal to 2), indicating that the edge offset type is used. If X is equal to Y, the syntax element is read as "sao_eo_class_luma" and if X is set equal to U and V, the syntax element is read as "sao_eo_class_chroma". Therefore, the edge offset class (corresponding to directions 0, 45, 90, and 135 degrees) is extracted from the bit stream in step 3105.

When the four offsets have been extracted, the process proceeds to step 3116, where it is tested whether the current color component X corresponds to the luminance component. If yes, test whether the edge offset sign method eo_sign_method_X (3114), that is, the index of the N value (in the conversion table "group") corresponding to the current color component (luminance) is greater than the threshold Th (3111). The threshold can be 0 or higher. If yes, the four symbols (3112) are extracted from the bit stream (3113). Then, the reading of the SAO parameters is completed and the procedure proceeds to step 3108.

Otherwise (edge offset sign method, ie, the index corresponding to the N value index is lower than the threshold), the sign is implicit and set as shown in FIG. 27B. Then, the reading of the SAO parameters is completed and the process proceeds to step 3108.

If the current color component is a chroma component, the process directly proceeds to step 3108.

This embodiment provides increased coding efficiency compared to the previous embodiment. This is because for chroma components, implicit signaling is often more efficient than explicit signaling of symbols.

FIG. 32 is a flowchart illustrating program steps for decoding SAO parameters according to the seventh embodiment of the second aspect.

According to this seventh embodiment, the sign of the edge shift is explicit (signaling) for peaks and valleys, and implicit for the other two groups when the color component is brightness. At this time, the symbol is transmitted for offsets O1 and O4, as shown in FIG. 27B.

In the initial step 3202, the "sao_type_idx_X" syntax element is read and decoded. The codeword representing this syntax element can use a fixed-length code or can use any arithmetic coding method. The syntax element sao_type_idx_X completes the determination of the type of SAO to be applied to the frame area for U and V processing of the color component Y or the two-chroma component.

In addition, although YUV color components are used for HEVC (sometimes called Y, Cr, and Cb components), it should be understood that other video encoding schemes, other color components, such as RGB color components, can also be used. The technique of the present invention is not limited to use with YUV color components, but can also be used with RGB color components or any other color components.

In the same step 3202, a test is performed to determine whether "sao_type_idx_X" is strictly positive. If "sao_type_idx_X" is equal to "0", it means that if X is set equal to Y, then there is no SAO for this frame area (CTU) of Y, and if X is set equal to U and V, then this frame area for U and V There is no SAO. The decision of the SAO parameter is completed and the process proceeds to step 3208. Otherwise, if "sao_type_idx" is strictly positive, it means that there is a SAO parameter for this CTU in the bit stream.

Then, the process proceeds to step 3206, where the loop performs four iterations. Four iterations can be performed in step 3207, where the absolute value of offset j is read and decoded by the bitstream. These four iterations correspond to the four absolute values of the four index offsets (O1, O2, O3, O4) of the SAO edge offset (see Figures 27A and 27B) or correspond to the four related to the SAO band offset Four absolute values of the range of offsets (see Figure 8).

After step 3207, the process proceeds to step 3203, where a test is performed to determine whether the type of SAO corresponds to the band offset type (sao_type_idx_X==1).

If the result is positive, the symbol for the offset with offset mode is decoded in steps 3209 and 3210 before the following step 3204 is executed, except for each offset with a value of zero, to read the bit stream And the position "sao_band_position_X" of the decoded SAO band is as shown in FIG. 8.

If the result in step 3203 is negative ("Sao_type_idx_X" is set equal to 2), indicating that the edge offset type is used. If X equals Y, read the syntax element as "sao_eo_class_luma" , And if X is set equal to U and V, the read syntax element is "sao_eo_class_chroma". Therefore, the edge offset class (corresponding to directions 0, 45, 90, and 135 degrees) is extracted from the bit stream in step 3205.

When the four offsets have been extracted, the process proceeds to step 3216, where it is tested whether the current color component X corresponds to the luminance color score. If so, for each of the four offsets (3212), decide whether the offset is a peak (j==1) or valley (j==4) offset, ie O1 or O4 (3215). If so, the offset sign is extracted from the bit stream (3213). Otherwise, the symbol is implicit and set as shown in Figure 27B. Then, the reading of the SAO parameters is completed and the process proceeds to step 3208. If the current color component is a chroma component, the process directly proceeds to step 3208.

The advantage of this embodiment is that it can be considered better for coding efficiency, because implicit signaling of chroma components is often more efficient than explicit signaling of symbols.

In a variation, an additional test can be performed to test whether the edge sign method, ie (in the "group" of the conversion table) the index corresponding to the N value is greater than the threshold Th.

In another embodiment, the signaling of all color components may allow simplification of hardware implementation.

Third group of embodiments (third aspect)

According to a third aspect of the present invention, a method for obtaining an edge offset is provided for performing sample adaptive offset (SAO) filtering, wherein when the symbol of the SAO offset is explicitly signaled in the bitstream, Its value is predicted.

In other words, the third aspect provides a method of performing sample adaptive offset (SAO) filtering on an image. The image includes a plurality of image parts. The method includes: Get most edge offsets to perform SAO filtering; Where at least one edge-offset symbol of the majority of edge offsets is sent in the bitstream; and Obtaining the majority of edge offsets includes predicting at least one edge offset whose symbol is sent to the bitstream. In fact, according to the embodiment of the third aspect, the offset value or more precisely the absolute offset value may be encoded with a variable length code or an arithmetically-encoded fixed code.

FIG. 33 illustrates the prediction procedure steps, which can be performed on the decoder side in the first embodiment of the third aspect in the (SAO) filtering embodiment. These steps can be performed during the procedure shown in FIG. For example, this prediction can be applied in step 701 of FIG. 11. These steps can be performed during the procedures shown in FIGS. 29 to 32, before the individual steps 2908 to 3208.

According to this first embodiment, at least one edge offset is predicted from a preset value. In this example, the edge offset is predicted based on the prediction sub-table that contains preset values. The advantage of this embodiment may be increased coding efficiency.

At the beginning of the program, a decoding offset table is obtained (3301). This table contains offset values with their individual symbols. The four offset values (3302) are then predicted (3303). This is done by adding the variable Offset_X_j of the component X (YUV) to the predictor Predictor_X_j using the predictor table (3304). The predictor table 3304 may contain different predictors applicable to various types of offsets (peaks, valleys, etc.). Therefore, on the encoder side, the predictor Predictor_X_j minus Offset_X_j value. The program then proceeds to step 3305.

According to an embodiment, the preset value is predetermined for at least one sequence, for example, all sequences, that is, the value of the prediction sub-table may be the same for the entire slice/picture frame or for all the frames of the sequence or all sequences.

According to an embodiment, a table of preset values is transmitted in the bit stream. For example, these preset predictor values are transmitted in the slice header. In a variation, these predetermined predictor values are transmitted in sequence parameters.

Due to better flexibility, these embodiments provide better coding efficiency.

According to an embodiment, these predetermined predictor values are predetermined for all sequences.

According to an embodiment, the default value symbol used to predict the edge offset is set, ie it is fixed. For example, the predictor values for offsets O1 and O2 are positive, and the predictor values for offsets O1 and O3 are negative.

According to an embodiment, the absolute predictor values for the edge offsets of the first and last types O1 and O4 are higher than the absolute predictor values for the edge offsets of other types O2 and O3.

This embodiment can be quite advantageous because the absolute average of offsets O1 and O4 is often greater than the absolute average of offsets O2 and O3. Therefore, this embodiment can provide better coding efficiency.

According to an embodiment, the absolute prediction sub-values for the first and last types (O1 and O4) of the edge offset are the same, and there are absolute predictions for the other types (O2 and O3) of the edge offset (O2 and O3) The sub-values are the same.

Preferably, the procedure is simplified and consistent with the statistical selection of offset values.

FIG. 34 illustrates a prediction procedure which can be executed on the decoder side according to the second embodiment of the third aspect during the SAO filtering embodiment. As mentioned above, these steps can be performed in the procedure shown in FIG. 10 or 11.

According to the second embodiment, the preset value depends on the sign parameter N and is used in the sign function to determine the index of the edge offset.

At the beginning of the program, a decoding offset table is obtained (3401). This table contains offset values with their individual symbols. The four offset values (3402) are then predicted (3403). This is done based on the predictor Predictor_X using the prediction sub-table (3404) and the formula group [eo_sign_method_X], which is based on the eo_sign_method_X group table (3406). The predictor table 3404 may contain different predictors, which are applicable to various types of offsets (peaks, valleys, etc.). At the same time, the table group (3406) is used to obtain the correct predictor for the current Offset_X_j. In this example, the prediction sub-table has a value N for its first dimension. But if the table group is the same regardless of the sequence, there is no need to convert eo_sign_method_X to the value N according to the table group. In addition, if eo_sign_method_X is not a direct N value, then the table "group" can be used to convert eo_sign_method to the correct N value.

Correspondingly, on the encoder side, the correct predictor Predictor_X_j is subtracted from the Offset_X_j value. The program then proceeds to step 3405.

The advantage of this embodiment may be an increase in coding efficiency compared to a fixed value for all edge offset sign methods. This is because when the N value of the symbol method increases, the absolute offset value increases.

According to an embodiment, at least some edge offsets of the majority of edge offsets whose symbols are signaled in the bitstream are not predicted, that is, only some offsets can be predicted and unpredicted offsets may have their signs displayed. Letter. Figures 35, 36, 37 and 38 show four examples of these embodiments. In these examples, these steps are similar to the procedures shown in Figure 33 and/or Figure 34, but they are only performed for partial offsets.

FIG. 35 illustrates the steps of the prediction procedure, which can be performed during the (SAO) filtering embodiment of the third embodiment according to the third aspect on the decoder side. As mentioned earlier, these steps can be performed during the procedure shown in FIG. 10 or 11.

According to this third embodiment, at least one edge offset is predicted by the symbol parameter N for the sign function used to determine the index of the edge offset. In addition, only these offsets are predicted.

For example, the offsets O1 and O4 are predicted by the predictor, and their absolute values are equal to the symbolic value N.

At the beginning of the program, a decoding offset table is obtained (3501). This table contains offset values with their individual symbols. For each offset (3502), it determines that the offset is a peak or valley offset, namely O1 or O4 (3507). If yes, the offset is the formula group [eo_sign_method_X] according to Predicted by the value (3503) obtained from the eo_sign_method_X group table (3506). According to the table shown in FIG. 7B, this value N is only the absolute value of the predictor, and if it is offset O1, the sign is positive, and if it is offset O4, the sign is negative. The process then proceeds to step 3505.

The advantages of this embodiment can increase coding efficiency.

According to other embodiments, other possible values for the absolute value of the predictor can also be considered as N-1 or N-2 or N+1 or N+2, especially if the edge offset sign method is used for the entire frame or The sequence is the same (if there is no contention between the edge offset sign methods).

FIG. 36 illustrates a prediction procedure which can be executed during the (SAO) filtering embodiment on the decoder side according to the fourth embodiment of the third aspect.

In this fourth embodiment, only the peak and valley shifts (O1 and O4) can be predicted. Therefore, this procedure is similar to the procedure shown in Figure 33, but only for partial offsets.

At the beginning of the program, a decoding offset table is obtained (3601). This table contains offset values with their individual symbols. For each offset (3602), decide whether the offset is a peak or valley offset, ie O1 or O4 (3607). If it is, the variable Offset_X_j of the component X (YUV) is added to the predictor Predictor_X_j (3603) using the predictor table (3604). The program then proceeds to step 3605.

The advantage of this embodiment may be increased coding efficiency. This is because the absolute average values of the offsets O1 and O4 are larger on average than those of the offsets O2 and O3. In addition, the absolute values used for O2 and O3 are often less than 1. Therefore, it may not be necessary to predict them and issue symbols.

FIG. 37 illustrates a prediction procedure that can be executed during the (SAO) filtering embodiment on the decoder side according to the fifth embodiment of the third aspect.

In this embodiment, the offset used for the luminance component is only prediction.

At the beginning of the program, the decoding offset table is obtained (3701). This table contains offset values with their individual symbols. For each offset (3702), it is determined whether the current color component is brightness (3708). If yes, the offset is predicted by adding the variable Offset_X_j of component X (YUV) to the predictor Predictor_X_j (3703) using the predictor table (3704). The program then proceeds to step 3705.

The advantage of this embodiment may be increased coding efficiency. This is because the absolute offset value is usually higher for the luminance component and lower for the chrominance component. The prediction of the luminance component can be sufficiently high. This embodiment can therefore provide simplification when the prediction of chroma shift is dispensed with.

FIG. 38 illustrates a prediction procedure step, which can be executed during the (SAO) filtering embodiment on the decoder side according to the sixth embodiment of the third aspect.

In this sixth embodiment, only the offset for the luminance component and the offsets O1 and O4 are predicted.

At the beginning of the program, a decoding offset table is obtained (3801). This table contains offset values with their individual symbols. For each offset (3802), it determines whether the current color component is brightness (3808). If it is, it is determined that the shift is a peak or valley shift, namely O1 or O4 (3807). If so, the offset is predicted by adding the variable offset_X_j of the component X (YUV) to the predictor Predictor_X_j (3803) using the predictor table (3804). The program then proceeds to step 3805.

According to other embodiments, at least one edge offset is predicted from the adjacent edge offset value, and only the sign of the adjacent edge offset is determined using the same sign method. In other words, if the previously offset symbol is determined using the same edge offset symbol method, at least a portion of the current edge offset value can be predicted based on the edge offset value of one of the previously encoded CTU parameters.

Fourth group of embodiments (fourth aspect)

According to a fourth aspect of the present invention, there is provided a method for obtaining an edge offset to perform sampling adaptive offset (SAO) filtering, wherein the sign of the at least one edge offset is implicit and the edge offset The minimum value is set to a preset value that is not zero.

In other words, the fourth aspect provides a method of performing sample adaptive offset (SAO) filtering on an image including most image parts. The method includes: Obtain most edge offsets to perform SAO filtering; Wherein obtaining the majority of edge offsets includes predicting at least one edge offset; and The sign of the at least one edge offset is implicit and the minimum value of the edge offset is set to a preset value that is not zero.

According to an embodiment of the fourth aspect, the offset used for edge offset classification has a minimum defined absolute value. In this embodiment, for an offset with a minimum defined absolute value greater than 0, the sign of the offset is not signaled in the bitstream.

It can be noted that in the current implementation of SAO, the edge offset is delimited as follows: O1＞=0 O2＞=0 O3＜=0 O4＜=0

According to the proposed embodiment, the offsets O1, O2, O3, and O4 have minimum absolute values MAO1, MAO2, MAO3, and MAO4, respectively. For example, the offset values O1, O2, O3, O4 are delimited as follows: O1＞=MAO1 O2＞=MAO2 O3＜=-MAO3 O4＜=-MAO4

The goal of this delimitation is to define the code rate for the offset. As in the current SAO implementation method, the absolute offset value Offset_X_j (607) shown in FIG. 10 can always be encoded from 0 to the maximum value. Therefore, on the encoder side, the absolute value Offset_X_j can be determined by MAO1_X, Predicted by MAO2_X, -MAO3_X, -MAO4_X.

On the decoder side, the same procedure as shown in FIG. 33 is applied to the predictor Predictor_X_j which can be set as follows: Predictor_X_1=MAO1 Predictor_X_2=MAO2 Predictor_X_3=-MAO3 Predictor_X_4=-MAO4

The advantage of this embodiment may be an increase in coding efficiency compared to the prior art. Since it does not need to add the offset symbols as a syntax element, this embodiment can be regarded as a simplification relative to the embodiment where the symbols are transmitted.

FIG. 39 shows systems 191 and 195, which include at least one of encoder 150 or decoder 100 and communication network 199 according to an embodiment of the present invention. According to an embodiment, the system 195 is used to process and provide content (for example, video and audio content for display/output or streaming video/audio content) to a user who has used the decoder 100 for example The user interface of a user terminal or a user terminal that can communicate with the decoder 100 accesses the decoder 100. The user terminal may be a computer, a mobile phone, a tablet computer, or any type of device, which can provide/display (promote/stream) content to the user. The system 195 acquires/receives the bit stream 101 via the communication network 199 (in the form of a continuous stream or signal, for example, while the previous video/audio system is being displayed/output). According to an embodiment, the system 191 is used to process content and store processed content, such as processed video and audio content, for subsequent display/output/streaming. The system 191 obtains/receives the content including the original sequence of the image 151, which is received and processed by the encoder 150 (including filtering with the deblocking filter according to the present invention), and the encoder 150 generates the bit stream 101, which is It is transmitted to the decoder 100 via the communication network 191. The bit stream 101 is then sent to the decoder 100 in several ways. For example, it can be generated in advance by the encoder 150 and stored in the storage device of the communication network 199 (for example, server or cloud storage) as data until the user Until the content from the storage device (ie, bit stream data) is requested, the data is transmitted/streamed from the storage device to the decoder 100 at the storage device point. The system 191 may also include a content providing device for providing/streaming content information of the content stored in the storage device (eg, content name and other original/storage location data for identification, selection, and requesting the content) Users (for example, data sent to the user interface will be displayed on the user terminal), and user requests for receiving and processing content, so that the requested content can be transmitted/streamed from the storage device to the user terminal . Alternatively, the encoder 150 generates a bit stream 101 and directly transmits/streams the content to the decoder 100 when the user requests the content. The decoder 100 then receives the bit stream 101 (or signal) and performs filtering with the deblocking filter according to the present invention to obtain/generate the video signal 109 and/or audio signal, which is then used by the user terminal to provide all Request content to users.

In the previous embodiment, the functions can be implemented in hardware, software, firmware, or a combination thereof. If implemented in software, the functions can be stored or transmitted as one or more instructions or codes on a computer-readable medium and executed by the hardware as the main processing unit.

The computer-readable medium may include a computer-readable storage medium, which corresponds to a tangible medium, such as a data storage medium, or a communication medium including any medium that can cause a computer program to be transferred from one place to another according to a communication protocol, for example. In this way, computer-readable media generally corresponds to (1) tangible computer-readable storage media, which is non-transitory, or (2) communication media, such as signals or carrier waves. The data storage medium may be any available medium, which may be accessed by one or more computers or one or more processors to obtain instructions, codes, and/or data structures for the implementation of the technology described in this case. The computer program product may include computer readable media.

For example, but not limited to, the computer-readable storage medium may include RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage, or other magnetic storage devices, flash memory, or any other media, which It can be used to store the desired code in the form of commands or data structures and can be accessed by the computer. In addition, any connection is preferably called a computer-readable medium. For example, if the command is transmitted by a website, server, or other remote source using coaxial cable, optical cable, twisted pair, digital subscriber line (DSL), or wireless technology such as infrared, radio, and microwave, then the coaxial cable, optical cable , Twisted pair, DSL, or wireless technologies such as infrared, radio, and microwave are included in the definition of media. However, it should be understood that computer-readable storage media and data storage media do not include connections, carrier waves, signals, or other transient media, but non-transient, tangible storage media. The discs used here include compact discs (CDs), laser discs, optical discs, digital versatile discs (DVDs), floppy discs, and Blu-ray discs, where discs usually play data magnetically, while lasers play data optically . The combination of these should also be included in the scope of computer-readable media.

The instructions may be, for example, one or more multi-bit signal processors (DSPs), general-purpose microprocessors, application specific integrated circuits (ASICs), field programmable logic arrays (FPGAs), or other equivalent integrated or discrete logic circuits. Or more processors. Therefore, the term "processor" as used herein may refer to any of the foregoing structure or any other structure suitable for implementation of the techniques described herein. In addition, in some aspects, the functions described herein may be provided in dedicated hardware and/or software modules, which are configured to encode or decode or incorporate into a combined codec. In addition, the technology may be fully implemented in one or more computers or logic units.

The technology in this case can be implemented in various devices or equipment, including wireless handsets, integrated circuits (ICs), or IC groups (eg, chipsets). The various components, modules or units described in this case are used to emphasize the functional aspects of the device, which are configured to perform the technology of this case, but do not necessarily need to be implemented by different hardware units. Conversely, as described above, various units can be combined into a codec hardware unit or provided by a collection of interactive hardware units that include one or more processors as described above with appropriate software and/or firmware.

Although the present invention has been shown and described in detail in the drawings and the foregoing description, these illustrations and descriptions are considered to be examples or examples and are not limiting, and the invention is not limited to the disclosed embodiments. Other changes to the disclosed embodiments can still be understood and acted upon by those skilled in the art by studying the drawings, the case, and the scope of the accompanying patent applications to substantiate (ie, execute) the claimed invention.

Some embodiments of the first, second, third and fourth aspects can be combined.

For example, symbols with offsets O1 and O4 (peaks and valleys) can be transmitted in the bitstream as luminance and chrominance components. For example, symbols with offsets O1 and O4 (peaks and valleys) can be transmitted in the bitstream as luminance and chrominance components. In addition, the absolute predictor value may be set equal to the value N of the sign method for offsetting O1 and O4 (peak and valley).

In another embodiment, the symbology can also be transmitted with a table set corresponding to the following set of N values: {0, 2, 4, 8}.

In the scope of the patent application, the use of "including" does not exclude other elements or steps, and the indefinite article "a" or "an" does not exclude the majority. A single processor or other unit may implement the functions of several elements described in the scope of the patent application. This means that different characteristics are described in different sub-items with each other. This does not mean that the combination of these characteristics cannot be used advantageously. The component symbols in the patent application scope should not be constructed to limit the scope of the present invention.

1‧‧‧Video sequence 2‧‧‧Image 3‧‧‧Slice 4‧‧‧ coding tree box 5‧‧‧Coding unit 6‧‧‧Prediction Unit 7‧‧‧ Conversion unit 200‧‧‧Data communication network 201‧‧‧Server 202‧‧‧Customer 204‧‧‧Data flow 300‧‧‧Processing device 302‧‧‧Communication interface 303‧‧‧Communication network 304‧‧‧Data storage method 305‧‧‧Drive 306‧‧‧disc 307‧‧‧Read-only memory 308‧‧‧Microphone 309‧‧‧ screen 310‧‧‧ keyboard 311‧‧‧Central Processing Unit 312‧‧‧random access memory 313‧‧‧Communication bus 320‧‧‧ digital camera 400‧‧‧Encoder 401‧‧‧ Original sequence image 402‧‧‧Cutting module 403‧‧‧Internal prediction module 404‧‧‧Motion estimation module 405‧‧‧Motion compensation module 406‧‧‧Select module 407‧‧‧ Conversion module 408‧‧‧Quantization module 409‧‧‧Entropy coding 410‧‧‧ bit stream 411‧‧‧ Dequantization module 412‧‧‧Inverse Conversion Module 413‧‧‧Internal prediction module 414‧‧‧Motion compensation module 415‧‧‧Post filter module 416‧‧‧Reference image/picture 417‧‧‧Motion vector prediction and coding module 418‧‧‧Action Vector Bar 50‧‧‧Loop filter program 51‧‧‧Initial steps 52‧‧‧deblocking filter 53‧‧‧Reconstruction 54‧‧‧Sample adaptive offset 55‧‧‧ frame 56‧‧‧Adaptive loop filter 57‧‧‧ Reconstruction 58‧‧‧Classification 60‧‧‧decoder 61‧‧‧ bit stream 62‧‧‧Entropy decoding 63‧‧‧ Dequantization 64‧‧‧Inverse conversion 65‧‧‧ forecast 66‧‧‧Motion compensation 67‧‧‧Post filtering 68‧‧‧Reference image/picture 69‧‧‧decode video signal 70‧‧‧Motion vector decoding module 71‧‧‧Action vector column data 40‧‧‧ group 41‧‧‧ belt 42‧‧‧belt 43‧‧‧ belt 44‧‧‧band

The embodiments of the present invention will now be described by way of illustration only and with reference to the following drawings.

FIG. 1 is a schematic diagram for explaining the coding structure used in HEVC;

FIG. 2 is a block diagram illustrating a data communication system in which one or more embodiments of the present invention can be implemented;

FIG. 3 is a block diagram illustrating elements of a processing device, in which one or more embodiments of the present invention may be implemented;

4 is a flowchart illustrating steps of an encoding method according to an embodiment of the invention;

FIG. 5 is a flowchart of the steps of the loop filtering process according to one or more embodiments of the present invention;

6 is a flowchart of steps of a decoding method according to an embodiment of the invention;

7A and 7B are diagrams for explaining edge-type SAO filtering in HEVC;

8 is a diagram for explaining band-type SAO filtering in HEVC;

FIG. 9 is a flowchart illustrating the procedure steps of decoding SAO parameters according to the HEVC specification;

10 is a flowchart illustrating a more detailed one step of the procedure of FIG. 9;

FIG. 11 is a flowchart illustrating how SAO filtering is performed on the video part according to the HEVC specification;

FIG. 12 is a flowchart of steps used to execute an encoder to determine SAO parameters for a group of CTUs (frames or slices);

13 is a flowchart for illustrating the steps of an example of statistics calculated on the encoder side that can be applied to an edge offset filter;

Figure 14 shows the steps of the first encoding procedure;

Figure 15 shows the steps of the second encoding procedure;

16 is a flowchart illustrating how SAO filtering is performed on the image portion according to the first embodiment of the first aspect of the present invention;

FIG. 17 illustrates the procedure of parsing the SAO parameters in the bit stream on the decoder side according to the first embodiment shown in FIG. 16;

18 illustrates a procedure of parsing SAO parameters in a bit stream on the decoder side according to the second embodiment;

FIG. 19 illustrates a procedure of parsing SAO parameters in a bit stream on the decoder side according to the third embodiment;

Fig. 20 illustrates an example of using an encoder to select eo_sign_method_X;

21 is a flowchart illustrating how the SAO filtering system is executed on the image portion according to the fourth embodiment of the first aspect of the present invention;

22 is a flowchart illustrating how the SAO filtering system is executed on the image portion according to the fifth embodiment of the first aspect of the present invention;

FIG. 23 illustrates a procedure of parsing the SAO parameters in the bit stream on the decoder side according to the sixth embodiment;

FIG. 24 is a flowchart illustrating the steps of an example in which statistics calculated on the encoder side are applied to an edge offset type filter;

FIG. 25 is a flowchart illustrating the steps of an encoder executed to determine the SAO parameters for a group of CTUs (frames or slices);

FIG. 26 is a flowchart illustrating a procedure of decoding SAO parameters according to the second embodiment of the second aspect;

27A and 27B are diagrams for explaining edge-type SAO filtering according to the embodiment;

28 is a flowchart of the procedure steps of decoding SAO parameters according to the third embodiment of the second aspect;

FIG. 29 is a flowchart of procedure steps of decoding SAO parameters according to the fourth embodiment of the second aspect;

30 is a flowchart of the procedure steps of decoding SAO parameters according to the fifth embodiment of the second aspect;

FIG. 31 is a flowchart of procedure steps of decoding SAO parameters according to the sixth embodiment of the second aspect;

32 is a flowchart of the procedure steps of decoding SAO parameters according to the seventh embodiment of the second aspect;

FIG. 33 illustrates steps of a prediction procedure that can be performed during the SAO filtering embodiment on the decoder side according to the first embodiment of the third aspect;

FIG. 34 illustrates steps of a prediction procedure that can be performed during the SAO filtering embodiment on the decoder side according to the second embodiment of the third aspect;

FIG. 35 illustrates steps of a prediction procedure that can be performed during the SAO filtering embodiment on the decoder side according to the third embodiment of the third aspect;

FIG. 36 illustrates steps of a prediction procedure that can be performed during the SAO filtering embodiment on the decoder side according to the fourth embodiment of the third aspect;

FIG. 37 illustrates steps of a prediction procedure that can be performed during the SAO filtering embodiment on the decoder side according to the fifth embodiment of the third aspect;

FIG. 38 illustrates steps of a prediction procedure that can be performed during the SAO filtering embodiment on the decoder side according to the sixth embodiment of the third aspect;

39 is a schematic diagram of a system including an encoder or decoder and a communication network according to an embodiment of the invention.

Claims (52)

A method for performing sample adaptive offset (SAO) filtering on an image including most image parts, the method includes: Get the edge offset to perform SAO filtering; Applying a predetermined symbol method to determine the index of the edge offset, applying the edge offset classification to at least one pixel of the image; The predetermined symbol method is sent in the bit stream. For example, in the method of claim 1, the predetermined symbol method is sent to the bit stream in CTU, frame, slice or sequence level. For example, the method of claim 1 or 2, wherein the predetermined symbol method is predicted. A method as claimed in any one of the aforementioned patent applications, wherein the predetermined symbolic method consideration corresponds to the number N of defined values to define a symbolic function. For example, in the method of claim 4, the predetermined symbol method is selected on the encoder side according to the discontinuous majority limit value N. For example, the method of claim 4 of the patent application, wherein the predetermined symbol method is selected on the encoder side according to the majority limit value N depending on the slice type. For example, the method of claim 4 of the patent application, wherein the predetermined symbol method is selected on the encoder side according to the majority limit value N depending on the color component. For example, the method of claim 4 of the patent application, wherein the predetermined symbol method is selected on the encoder side according to the majority limit value N that depends only on the luminance component. For example, the method of claim 4 of the patent application, wherein the predetermined symbol method is selected on the encoder side according to the majority limit value N in the slice header transmitted. For example, in the method of claim 4, the predetermined symbol method is selected on the encoder side according to the majority limit value N. A method for providing edge offset to perform sample adaptive offset (SAO) filtering on an image, the method includes: Calculate the edge offset to perform the SAO filtering; Choose one of the most possible symbologies to use to determine the index of the edge offset; and The selected symbol method is signaled in the bitstream, thereby allowing the edge offset classification to be performed according to the selected symbol method. For example, the method of claim 11 of the patent scope, wherein the symbol method is sent in the bit stream in CTU, frame, slice or sequence level. For example, the method of claim 11 or 12, wherein the symbol method is predicted. For example, the method of any one of the items 11 to 13 of the patent application scope, wherein the symbol method is selected according to the parameters of the current slice or frame. The method as in any one of the items 11 to 14 of the patent application scope, wherein the symbolic method considers the number N corresponding to the limit value to define a symbolic function. A method for performing sample adaptive offset (SAO) filtering on an image including most image parts, the method includes: Obtain most edge offsets to perform SAO filtering; At least one edge-shifted symbol of the majority of edge shifts is sent in the bit stream. As in the method of claim 16, the symbol of the edge offset is sent to all edge offsets of the majority of edge offsets, where the symbol function used to determine the index of the edge offset is used in The sign parameter N is positive. For example, the method of claim 16 of the patent application, in which the symbol of the edge offset is sent to all edge offsets of the majority of edge offsets, if used in the symbol function used to determine the index of the edge offset If all the symbol parameters N are higher than the predetermined threshold. For example, in the method of claim 16, wherein in the at least one edge offset, the symbol whose signal is sent belongs to the peak and valley structure. For example, in the method of claim 16 of the patent application range, regardless of the predetermined symbol method used to determine the index of the edge offset, the symbol of the at least one edge offset is sent. For example, in the method of claim 16, the symbol of the edge offset is only sent to the edge offset, and the symbol parameter N used to determine the index of the edge offset is higher than a predetermined threshold. For example, in the method of claim 16, the symbol of the edge offset is only sent to the edge offset, and the symbol parameter N used to determine the indexes of the edge offsets is not zero. A method as in any one of claims 16 to 22, wherein the symbol with the edge shifted is sent only for the luminance component. A method for providing edge offset to perform sample adaptive offset (SAO) filtering on an image, the method includes: Obtain the majority of edge offsets to perform the SAO filtering; Calculate the index of the edge offset by applying the sign method to the symbol parameter N for each edge offset; and At least one edge-offset symbol of the majority of edge offsets is signaled in the bit stream. A method for performing sample adaptive offset (SAO) filtering on an image including most image parts, the method includes: Obtain most edge offsets to perform SAO filtering; Where at least one edge-shifted symbol of the majority of edge shifts is signaled in the bitstream; and Obtaining the majority of edge offsets includes predicting at least one edge offset whose symbol is sent to the bitstream. As in the method of claim 25, wherein the at least one edge offset is predicted from a preset value. For example, the method of claim 26, wherein the preset value is transmitted in the bit stream. For example, the method of claim 26, wherein the preset value is predetermined for at least one sequence. For example, the method of applying for item 26 of the patent scope further includes setting the symbol of the preset value. For example, the method of claim 26, wherein the preset value depends on the symbol parameter N, which is used in the symbol function used to determine the index of the edge offset. For example, the method of applying for item 25 of the patent scope further includes: Applying edge offset classification to pixels of the image to classify the pixel into four types, each type being associated with a given edge offset, and Among them, the absolute predictor values for the first and last types of edge shifts corresponding to the peaks and valleys are higher than those for other types of edge shifts. For example, the method of applying for item 25 of the patent scope further includes: Applying edge offset classification to pixels of the image to classify the pixel into four types, each type having a given edge offset, and There is the same absolute predictor value for the edge shifts of the first and last types corresponding to the peaks and valleys. For example, in the method of claim 32, there are other types of edge predictors with the same absolute predictor value. For example, the method of claim 31 or 32, wherein the edge offset classification considers at least one edge offset of the majority of edge offsets to be transmitted to the symbol in the bitstream. For example, the method of claim 25, wherein the at least one edge offset is predicted by a symbol parameter N, and the symbol parameter N is used in the symbol function used to determine the index of the edge offset. For example, the method of claim 25, wherein in the majority of edge offsets, at least some edge offsets whose symbols are signaled in the bitstream are not predicted. For example, the method of applying for item 25 of the patent scope further includes: Apply an edge offset to the pixels of the image to classify the pixel into four types, each type is related to a given edge offset, and Only the first and last types of edge shifts corresponding to the peaks and valleys are predicted. As in the method of claim 25, where the edge shift is only predicted for the luminance component. For example, the method of applying for item 25 of the patent scope further includes: Apply an edge offset to the pixels of the image to classify the pixel into four types, each type is related to a given edge offset, and Among them, only the first and last types of the edge shift corresponding to the peaks and valleys and the edge shift for the luminance component are predicted. For example, the method of claim 25, wherein the at least one edge offset is predicted from the adjacent edge offset value. For example, in the method of claim 25, wherein only when the adjacent edge offset is determined using the same sign method, the at least one edge offset is predicted from the adjacent edge offset value. A method for providing edge offset to perform sample adaptive offset (SAO) filtering on an image, the method includes: Obtain the majority of edge offsets to perform the SAO filtering; For each edge offset, obtain the sign of the edge offset; The method further includes signaling the symbol of at least one edge offset of the majority of edge offsets in the bitstream; and Obtaining the majority of edge offsets includes predicting at least one edge offset whose symbol is sent to the bitstream. A method for performing sample adaptive offset (SAO) filtering on an image including most image parts, the method includes: Obtain most edge offsets to perform SAO filtering; Wherein obtaining the majority of edge offsets includes predicting at least one edge offset; and The sign of the at least one edge offset is implicit and the minimum value of the edge offset is set to a preset value that is not zero. A method for providing edge offset to perform sample adaptive offset (SAO) filtering on an image, the method includes: Calculate the majority of edge offsets to perform the SAO filtering; Where calculating most edge offsets includes predicting at least one edge offset, and The sign of the at least one edge offset is implicit and the minimum value of the edge offset is set to a preset value that is not zero. A method for encoding an image, including using any one of the patent application items 11, 24, 42 or 44 to perform sample adaptive offset (SAO) filtering. A method of decoding an image, including using the method of any one of patent application items 1, 16, 25, or 43, to perform sample adaptive offset (SAO) filtering. An apparatus for performing sample adaptive offset (SAO) filtering on an image including a majority of image parts. The apparatus includes a microprocessor and is structured to perform any one of items 11, 24, 42, or 44 of the patent application. Method steps. An apparatus for performing sample adaptive offset (SAO) filtering on an image including a majority of image parts. The apparatus includes a microprocessor and is structured to perform any one of items 1, 16, 25, or 43 of the patent application. Method steps. An encoder including the device as claimed in item 46 of the patent scope. A decoder including a device as claimed in item 47 of the patent scope. A program, when executed by a computer or a processor, causes the computer or the processor to execute the method of any one of items 1, 11, 16, 24, 25, 42, 43, and 44 of the patent application. A system including an encoder as claimed in item 49 and a decoder as claimed in item 50 is constructed to communicate through a communication network.

Images