JP3312417B2 - Image signal encoding device and image signal decoding device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image signal encoding apparatus and an image signal decoding apparatus for encoding a moving image signal.

[0002]

2. Description of the Related Art When a moving image is digitized and recorded and transmitted, the data is coded (compressed) because the amount of data becomes enormous. As a typical encoding method, there is Moving Picture Expert Group (MPEG) 1. MPEG1 is an ISO (International Organization for Standardization) and IE
C (Joint TechnicalCom)
mittee) 1 SC (Sub Committee) 29 WG (Worki)
ng Group 11). In MPEG1, a hybrid (Hybrid) system in which motion compensation prediction coding and DCT (Discrete Cosine Transform) coding are combined is adopted.

[0003] The motion compensation prediction coding is a method using the correlation in the time axis direction of an image signal, and predicts a currently input image from an image signal which has already been decoded and reproduced, and performs the decoding and reproduction. This method compresses the amount of information necessary for encoding by moving an image signal that is known in accordance with the motion of the signal and transmitting the motion information (motion prediction vector) at that time and the prediction error at that time. In MPEG1, one image (picture) is divided into small blocks (called MBs and macroblocks and is composed of 16 lines × 16 pixels).
And motion compensation prediction coding is performed in units of small blocks.

The motion compensation prediction error signal at this time is subjected to DCT. DCT coding uses two-dimensional correlation in an image of an image signal to concentrate signal power on a specific frequency component, and encodes only the concentratedly distributed coefficients to compress the amount of information. Make it possible. In MPEG1, DC
T is applied to each block composed of 8 lines × 8 pixels.

FIG. 1 shows the relationship between MBs and blocks in MPEG1. In MPEG1, since the image format is a component signal of the 4: 2: 0 format, the MB is composed of four luminance blocks adjacent to the left, right, up and down, and Cb and Cr color difference blocks at the same position on the image. It is composed of a total of six blocks.
The order of transmission is Y0, Y1, Y2, Y3, Cb, Cr.

In MPEG1, when transmitting a motion compensation prediction error signal of this MB layer, a CBP (Coded Block Pattern) indicating whether or not six blocks in the MB have non-zero DCT coefficients to be transmitted. Variable length code called
(VLC, Variable LengthCode) is added to the header of the MB layer and transmitted. The CBP exists if at least one of the six blocks in the MB has a non-zero coefficient.

FIG. 4 shows a VLC table at this time. In the order of the blocks (Y0, Y1, Y2, Y3, Cb, Cr), "1" with non-zero coefficient and "0" without non-zero coefficient for each block.
In this case, the values when arranged in binary from the MSB (Most Significant Bit) are used as CBP values, and VLC codes corresponding to the respective values are given. In the decoding, the VLC is converted to binary notation as shown in FIG. 4, and there is a non-zero coefficient in a block which becomes "1" in the order of (Y0, Y1, Y2, Y3, Cb, Cr) from the MSB. become.

For example, the shortest VLC code of "111" indicates that non-zero coefficients exist only in the blocks Y0, Y1, Y2, Y3 (the non-zero coefficients are in the blocks Cb, Cr). not exist). In the configuration of the VLC table of the CBP for the component signal of the 4: 2: 0 format in MPEG1, a short VLC is allocated when a non-zero coefficient does not exist in the color difference signal block (Cb, Cr block).

In this VLC table, b is
The number before that is a binary number, and the number in parentheses is a code represented by a binary number represented by a decimal number.

[0010] In recent years, MPEG2 has been replaced by MPEG1.
In this regard, there has been studied a method of targeting not only a 4: 2: 0 format component signal but also a 4: 2: 2 format component signal and a 4: 4: 4 format component signal to an image signal to be encoded. . In particular, the 4: 2: 2 format of component signals is recommended by CCIR (International Advisory Committee on Radio Communications).
This format is widely known as endation 601 (Rec. 601), and is widely used as a recording format of an image signal used in a broadcasting station or the like.

FIGS. 2 and 3 show 4: 2: 2 and 4:
The relationship between the MB and the block in each case of the 4: 4 format component signal is shown. In the 4: 2: 2 format component signal, the MB is composed of eight blocks, that is, four luminance blocks adjacent to the left, right, up, and down and color difference blocks of Cb0, Cb1, Cr0, and Cr1 at the same position on the image. It consists of. The order of transmission is Y0, Y1, Y2, Y3, Cb0, Cr0, Cb1, Cr1.

In the 4: 4: 4 format component signal, the MB is composed of four luminance blocks adjacent to the left, right, up, and down, and Cb0, Cb1, Cb at the same position on the image.
Each of the color difference blocks b2, Cb3, Cr0, Cr1, Cr2, and Cr3 is composed of 12 blocks in total. The order of transmission is Y
0, Y1, Y2, Y3, Cb0, Cr0, Cb1, Cr1, Cb2, Cr2, Cb
3, Cr3.

[0013]

Problems to be Solved by the Invention 4: 2: 0 and 4:
One of the problems in encoding the component signal in the 2: 2 and 4: 4: 4 format is the handling of the CBP described above. This is because the number of blocks included in the MB differs between the respective formats, and the VLC table (FIG. 4) for the component signal of the 4: 2: 0 format in the conventional MPEG1 is set to 4:
The problem is that it cannot be used with component signals of the 2: 2 and 4: 4: 4 formats.

To solve the above problem, 4: 2: 0
And CBP (Coded Blope), which has a high coding efficiency for component image signals of 4: 2: 2 and 4: 4: 4 format and has a simple coding and decoding apparatus.
ck Pattern) code encoding technology is required.

In general, when a component image signal is encoded by a hybrid encoding method, a motion compensation prediction error signal in a macroblock has a non-zero coefficient only in a luminance signal block (Y block) and a chrominance signal block. In many cases, non-zero coefficients do not exist in (Cb, Cr blocks).

Therefore, in order to more efficiently encode and decode the CBP code, it is necessary to make good use of the above properties.

The present invention has been proposed in view of the above-described circumstances, and has the following aspects: (4: 2: 0), (4: 2: 2)
And an image signal decoding apparatus and an image signal decoding apparatus capable of efficiently encoding CBP for any component image signal in the (4: 4: 4) format. .

[0018]

SUMMARY OF THE INVENTION In order to achieve the above object, the present invention proposes an image signal encoding apparatus for encoding an input image signal. Divided into macroblocks consisting of pixels,
The image processing apparatus includes a compression unit that performs a predetermined conversion for a compression process on a per-macroblock basis and outputs a conversion coefficient, and a variable-length encoding unit that performs variable-length encoding on the conversion coefficient. In this image signal encoding apparatus, the variable length encoding means includes a C for indicating which of the small blocks into which the macroblock is further divided has a non-zero transform coefficient.
As a VLC table for performing variable-length encoding of a BP code, a different VLC table is provided for each of a luminance signal block and a chrominance signal block. 0), (4: 2:
Regardless of the component signal of the 2) and (4: 4: 4) formats, the CBP code of the color difference signal is subjected to variable length coding using the same VLC table for the color difference signal block, When the input image signal is in the (4: 2: 0) format, the CBP code of the color difference signal is arbitrarily encoded with 2 bits out of 4 bits, and the input image signal is in the (4: 4: 4) format. For the CBP code of the color difference signal, the VLC table for the color difference signal block is applied twice.

Further, according to the present invention, an image signal encoded in units of macroblocks obtained by dividing one screen into a plurality of frames is subjected to inverse variable-length encoding, and a compressed image signal and a small image obtained by further dividing the macroblocks are obtained. The CBP code for indicating in which of the blocks a non-zero transform coefficient exists is separated from the CBP code.
An image signal decoding apparatus for decoding the compressed image signal based on a BP code, a first inverse VLC table for inverse variable length encoding of a variable length encoded CBP code of a luminance signal block, and a color difference signal A second inverse VLC table for inverse variable length encoding of the variable length encoded CBP code of the block;
An inverse variable-length encoding unit that performs inverse variable-length encoding on the BP code and outputs a transform coefficient, and an expanding unit that expands the transform coefficient. In this image signal decoding apparatus, the inverse variable length encoding means may be configured to determine that the image signal is (4: 2: 0);
Regardless of the component signal of the (4: 2: 2) or (4: 4: 4) format, the same second inverse VL is used for the CBP code of the color difference signal block.
When inverse variable length encoding is performed using the C table and the image signal is in the (4: 2: 0) format, the CBP code of the color difference signal block is 2 out of 4 bits.
The bits are regarded as an arbitrary code, and the image signal is (4:
4: 4) In the case of the format, the second VLC table is applied twice for the CBP code of the color difference signal block.

That is, in the present invention, the CBP VL having different properties for the luminance signal block and the color difference signal block is used.
A C table is prepared, and the CBP of the block of the luminance signal and the chrominance signal in the macro block is encoded with reference to the respective VLC tables.

At this time, in order to simplify the encoding method,
4: 2: 0, 4: 2: 2, of component image signals
Regardless of the 4: 4: 4 format, the CBP VLC tables for the luminance signal block and the chrominance signal block are shared.

[0022]

According to the present invention, there is provided an image signal encoding apparatus and an image signal decoding apparatus, each of which has a component image signal of (4: 2: 0), (4: 2: 2) and (4: 4: 4) format. Since the luminance signal blocks in the macro block have the same configuration, the CB for the luminance signal block
P can be shared regardless of the format of the component image signal. Therefore, the VLC table and the inverse VLC table at the time of decoding can be reduced.

Also, the VL of the CBP for the color difference signal block
By sharing the C table in the 4: 2: 0 and 4: 2: 2 formats, in variable-length coding of CBP values, it is not necessary to use a table that distinguishes the difference between these two formats. Can be processed. still,
In the case of 4: 4: 4 format, CB for color difference signal is used.
The VLC of P is applied twice and efficiently processed.

[0024]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Preferred embodiments of the present invention will be described below with reference to the drawings.

An embodiment of a moving picture coding apparatus having CBP code coding means according to the present invention will be described with reference to FIG.

In the present encoding apparatus, the input image is
The encoding is performed based on the data structure in MPEG1 as shown in FIG. Each data layer will be briefly described below. 1. Block Layer A block is composed of, for example, 8 lines × 8 pixels next to luminance or color difference. For example, DCT (Discrete C
The sine transform is performed in this unit. 2. MB (Macro Block) Layer The block configuration of the MB is as shown in FIGS. What is used for the motion compensation mode and whether or not the prediction error need not be sent are determined in this unit. 3. Slice layer Consists of one or more macroblocks connected in the scanning order of the image. At the beginning of a slice, the first macroblock has data indicating its position in the image, and is designed to be able to recover in the event of an error.
Therefore, the length of the slice and the starting position are arbitrary and can be changed depending on the error state of the transmission path. 4. Picture layer A picture, that is, each image is composed of at least one or a plurality of slices. Then, according to the encoding scheme, the picture is classified into I picture, P picture, and B picture. 5. GOP Layer A GOP is composed of one or more I-pictures and zero or more non-I-pictures. 6. Video Sequence Layer A video sequence is composed of one or a plurality of GOPs having the same image size, image rate, and the like.

Information for controlling the basic operation of the encoding apparatus is stored in the memory 18. These include an image frame size, an output bit rate of encoded information, a motion prediction compensation method, and the like. These pieces of information are output as S25.

A moving image to be encoded is input to an image input terminal 10.
Input from The input image signal is supplied to the field memory group 11. From the field memory group 11, the macroblock signal S1 to be currently encoded is supplied to the hybrid encoder 12.

The hybrid encoder 12 is a hybrid (hybrid) code that combines motion compensation predictive coding and transform coding such as DCT (Discrete Cosine Transform), which are typical examples of high-efficiency video coding. Is performed. The configuration does not affect the main subject of the present invention, and thus the description thereof is omitted here.

The motion-compensated prediction error signal S2 of the MB layer output from the hybrid encoder 12 is variable-length coded by a VLC (variable-length coder) 13 into a Huffman code or the like. At this time, CBP (Coded B) indicating whether a block in the MB has a non-zero DCT coefficient to be transmitted.
A variable length code called “lock pattern” is added to the header of the MB layer and transmitted. The CBP is transmitted if at least one block in the MB has a non-zero coefficient. The CBP receives the motion compensation prediction error signal S2 and is configured by a CBP constructor 16. In the CBP constructor 16, the CB
P is the value in the luminance signal block and C in the color difference signal block.
It is determined separately for the BP value.

The configuration of the MB luminance signal block is as follows:
The component image signals of the 2: 0, 4: 2: 2, 4: 4: 4 format are all the same, and the CBP is non-zero for each block in the order of (Y0, Y1, Y2, Y3). Assuming that the coefficient is “1” and the coefficient is “0”, a value obtained by arranging from a MSB (Most Significant Bit) and displaying it as a 4-bit binary number is defined as a CBP value.

The configuration of the MB color difference signal block is as follows:
The component image signals of the 2: 0, 4: 2: 2, 4: 4: 4 format are all different, but here, as in the case of the luminance signal block, the CBP of the chrominance signal block is 2 bits of 2 bits. Expressed in radix code. That is, in the 4: 2: 0 format, 4-bit coding is performed as (Cb, Cr, *, *). Here, "*" means Don't Care, that is, any code is acceptable. In 4: 2: 2 format,
(Cb0, Cr0, Cb1, Cr1) is coded as a 4-bit code. 4:
In the 4: 4 format, (Cb0, Cr0, Cb1, Cr1) and (Cb2, Cr
2, Cb3, Cr3).

With respect to the CBP value of the luminance signal block and the CBP value of the color difference signal block obtained in this way, VLC
The VLC is applied to each vessel 13. FIGS. 8 and 9 show the variable-length code tables at this time. Here, the CLC VLC tables for the luminance signal block and the chrominance signal block are shared regardless of the 4: 2: 0, 4: 2: 2, and 4: 4: 4 formats of the component image signals. Therefore, when the CBP is converted into a VLC, the VLC unit 13 has the following configuration:
There is no need to distinguish between the 2: 0, 4: 2: 2 formats. In the case of 4: 4: 4 format, CBP VLC for color difference signals is applied twice. Whether to apply twice is determined by the format of the input image signal. In the table,
The numbers in parentheses represent codes represented by binary numbers in decimal numbers.

Note that the luminance signal block C
BP VLC "000000" and color difference signal block CB
Since the combination of P and VLC "0" is meaningless, it is desirable to prohibit the generation of that pattern.

In the above example, the simplicity of the encoder is prioritized. However, if the highest priority is given to the CBP encoding efficiency in the 4: 2: 0 format, the 4: 2: 0 format shown in FIG. A CBP VLC table for a color difference signal block is prepared.
The table of FIG. 9 is used for 4: 2: 2 and 4: 4: 4 formats. As described above, CB
Construct P and convert them to VLC.

The variable length code output from the VLC unit 13 is stored in a buffer memory 14 and then output to an output terminal 15.
Transmits a bit stream at a constant transmission rate.

Next, a moving picture decoding apparatus corresponding to the above moving picture coding apparatus will be described with reference to FIG.

The bit stream signal input from the input terminal 50 is stored in a buffer memory 51 and then supplied to an inverse VLC unit 52.

As described in the description of the encoding apparatus, the bit stream is composed of six layers, that is, video sequence, GOP, picture, slice, macro block, and block. For the layers of the video sequence, the GOP, the picture, and the slice, a start code indicating that they are started is received at the head of each layer, and thereafter, header information for controlling decoding of an image is received. Upon receiving each start code, the inverse VLC unit 52 decodes the header information of each layer, and stores the obtained control information for image decoding in the memory 201. These pieces of information are output as S104.

The motion compensation prediction error signal S80 of the MB layer supplied from the inverse VLC unit is supplied to the hybrid decoder 53. The hybrid decoder 53 performs hybrid decoding combining motion compensation and transform coding such as inverse DCT (Invers Discrete Cosine Transform), which are typical examples of high-efficiency video coding. The configuration does not affect the main subject of the present invention, and thus the description thereof is omitted here.

At this time, the CBP VLC indicating which block in the MB has a non-zero DCT coefficient.
Is received in the MB layer header. 8 and 9 independently for the luminance signal block and the color difference block.
Is decrypted with reference to the table of FIG.

The structure of the MB luminance signal block is as follows:
In the respective component image signals of the 2: 0, 4: 2: 2, 4: 4: 4 format, they are all the same, and the CBP is viewed in the block order of (Y0, Y1, Y2, Y3).
There is a non-zero coefficient in the block that becomes 1 ".

The configuration of the color difference signal block of MB is 4:
Although each component image signal of the 2: 0, 4: 2: 2, 4: 4: 4 format is different, the CBP of the chrominance signal block is a 4-bit binary number as in the case of the luminance signal block. Obtained in units of code. That is, in the 4: 2: 0 format, a 4-bit code is decoded as (Cb, Cr, *, *). Here, "*" is Don't Care, that is, an arbitrary code.

In the 4: 2: 2 format, (Cb0, Cr0, C
b1, Cr1) is decoded as a 4-bit code. 4: 4: 4
The format is (Cb0, Cr0, Cb1, Cr1) and (Cb2, Cr2, Cb3,
Two 4-bit codes of Cr3) are decoded. CBP is
There is a non-zero coefficient in the block that is "1" when viewed from the first bit.

According to this method, regardless of the 4: 2: 0, 4: 2: 2, 4: 4: 4 format of the component image signal, the VLC table of the CBP for the luminance signal block and the CBP for the chrominance signal block is used. Can be shared.
Therefore, in decoding the CBP, the inverse VLC unit 52
There is no need to distinguish between the two formats between 4: 2: 0 and 4: 2: 2 formats. In addition,
In the case of 4: 4: 4 format, CB for color difference signal is used.
The inverse VLC of P is applied twice in succession.

When the encoder uses the VLC table shown in FIG. 10 to convert the CBP values for the 4: 2: 0 chrominance signal block into VLC, the inverse VLC unit also corresponds to the case. In this case, the table in FIG.
It is used for color difference signal blocks of 2: 2 and 4: 4: 4 format. As described above, the CBP is decoded.

Based on the CBP, decoded macroblock layer data S81 is output from terminal 55. As described above, the image data is restored from the bit stream data.

[0048]

According to the above-mentioned method, it is possible to efficiently transmit the CBP value for the component image signal of the 4: 2: 0, 4: 2: 2 and 4: 4: 4 format.

In each of the 4: 2: 0, 4: 2: 2, and 4: 4: 4 format component image signals, the luminance signal block in the macroblock has the same configuration, so the CBP for the luminance signal block is , And can be shared regardless of the format of the component image signal.

By sharing the VLC table of the CBP for the chrominance signal block in the 4: 2: 0 and 4: 2: 2 formats, the difference between these two formats in variable-length coding and decoding of CBP values is described. Need not be distinguished.

When the size of the variable length table of the CBP is MP
The number is reduced from 64 entries to 32 entries compared to EG1. Also, the maximum bit length of the variable length code is MPEG1.
Is reduced from 9-bit width to 7-bit width.

[Brief description of the drawings]

FIG. 1 is a diagram illustrating a block configuration of an MB in a 4: 2: 0 format signal.

FIG. 2 is a diagram illustrating a block configuration of an MB in a 4: 2: 2 format signal.

FIG. 3 is a diagram illustrating a block configuration of an MB in a 4: 4: 4 format signal.

FIG. 4 is a VLC table of CBP in MPEG1.

FIG. 5 is a diagram showing a data structure in MPEG1.

FIG. 6 is a block diagram of an encoder according to the present embodiment.

FIG. 7 is a block diagram of a decoder according to the embodiment.

FIG. 8 is a diagram showing V of CBP for a luminance signal block according to the present embodiment.
It is an LC table.

FIG. 9 is a diagram showing V of CBP for a color difference signal block according to the present embodiment.
It is an LC table.

FIG. 10 is a CLC VLC table for a color difference signal block dedicated to 4: 2: 0 format.

[Explanation of symbols]

11 field memory group, 12 hybrid encoder, 13 VLC unit (variable length encoder), 14
Buffer memory, 16 CBP constructor, memory,
51 buffer memory, 52 inverse VLC unit, hybrid decoder, 201 memory

Claims (2)

(57) [Claims] An image signal encoding apparatus for encoding an input image signal, wherein one screen of the input image signal is divided into macroblocks including a plurality of pixels, and each macroblock is used for compression processing. A compression unit that performs a predetermined conversion and outputs a transform coefficient; and a variable-length encoding unit that performs a variable-length encoding of the transform coefficient. The variable-length encoding unit includes a small block obtained by further dividing the macroblock. VL for performing variable-length coding on a CBP code for indicating where a non-zero transform coefficient exists.
As C table, and have a different VLC tables each luminance signal block and color difference signal block, the variable-length encoding means, said input image signal (4:
(2: 0), (4: 2: 2) and (4: 4: 4) formers
Color difference for any component signal
The same color difference signal block is used for the CBP code of the signal.
Variable length encoding is performed using the VLC table for locking.
The input image signal is in (4: 2: 0) format.
At one time, the CBP code of the color difference signal
2 bits of the input image signal are arbitrarily encoded.
Signal is in (4: 4: 4) format, the color difference signal
For the color difference signal block
An image signal encoding apparatus , wherein the VLC table is applied twice . 2. An image signal encoded in units of macroblocks, which is obtained by dividing one screen into a plurality of frames, is subjected to inverse variable-length encoding to generate a compressed image signal or a small block obtained by further dividing the macroblock. An image signal decoding device for separating a CBP code for indicating whether or not a non-zero transform coefficient exists and decoding the compressed image signal based on the CBP code. Using a first inverse VLC table for inverse variable length encoding of a CBP code and a second inverse VLC table for inverse variable length encoding of a variable length encoded CBP code of a color difference signal block An inverse variable length encoding unit that performs an inverse variable length encoding on the variable length encoded CBP code and outputs a transform coefficient; and an expansion unit that extends the transform coefficient. The row indicates that the image signal is (4: 2:
0), (4: 2: 2) and (4: 4: 4) formats, the same CBP code of the color difference signal block is obtained by using the same second inverse VLC table. When inverse variable length coding is performed and the image signal is in the (4: 2: 0) format, 2 bits out of 4 bits are regarded as an arbitrary code for the CBP code of the color difference signal block, Is an (4: 4: 4) format, the second VLC table is applied twice for the CBP code of the color difference signal block.