KR100338801B1 - digital data encoder/decoder method and apparatus - Google Patents

Abstract

PURPOSE: A method and apparatus of encoding and decoding digital data is provided to perform a decoding operation according to a descending the gravity of important degrees of digital data and additional information, in order to minimize deterioration of a quality with a part of a bit stream to restore data close to original data. CONSTITUTION: A decoding apparatus decodes encoded digital data according to a descending the gravity of important degrees of digital data bits. A bit stream analysis(400) part analyzes a bit important degree of the encoded digital data bit stream. A decoding part(410) decodes the analyzed digital data according to a descending the gravity of important degrees of data bits. A bit joining part(420) joins additional information with respect to each bit of each sample from data decoded in the decoding part(410).

Description

Digital data encoder / decoder method and apparatus

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to digital data encoding and decoding, and more particularly, to a digital data encoding / decoding method and apparatus for dividing digital data into digit units to encode and decode digital data in order of increasing importance.

In general, a waveform containing information is an analog signal that is continuous in time and continuous in time. Therefore, analog-to-digital (A / D) conversion is required to represent waveforms as discrete signals.

A / D conversion requires two steps. One is the sampling process of converting a continuous signal in time into a discrete signal. The other is an amplitude quantization process to limit the number of possible amplitudes to finite values. That is, quantization of amplitude is the process of converting input amplitude x (n) to y (n), an element of a finite set of possible amplitudes at time n.

If the quantized signal is simply encoded using a method such as PCM (Pulse Code Modulation), it is very convenient to use the quantized data because no other processing is required. However, this simple coding method is not optimal from the point of view of the size of data required to store or transmit data, even if the input samples are statistically independent, and even more insufficient if the input samples are statistically dependent.

Because of this problem, encoding is performed including lossless coding such as entropy coding or adaptive quantization of some kind. Therefore, the process is rather complicated than the simple method of storing only PCM data. The bitstream consists of additional information for compressing a signal as well as quantized data.

However, the information thus constructed generates a bitstream by encoding quantized data in sample units in order of data without considering importance. You can construct the bitstream regardless of its importance, as long as you simply save the bitstream and restore it again without any error.

When a bitstream is transmitted through a communication network, some bitstreams may be lost depending on the state of the communication network. In addition, when an error occurs during the transmission of the bitstream, the information of the bitstream after the error is propagated and restored to the wrong information. If only some bitstreams of the entire bitstreams need to be restored with the correct information, a higher quality signal may be restored than a bitstream in which the bitstream restored with the correct information is not restored, thereby preventing further quality degradation. .

In general, encoding techniques are given a fixed bit rate to the encoder, find an optimal state for the given bit rate, and quantize and encode to produce a bitstream according to the given bit rate. In the conventional coding scheme, the order of the bitstreams is configured with a size suitable for a given bit rate without much consideration.

In fact, when this configured bitstream is transmitted through a communication network, the bitstream is split into several slots. If the transmission line is overloaded or the bandwidth of the transmission line is too narrow, the entire slot sent by the transmitter cannot reach and only a part of the slot arrives at the receiver. In addition, deterioration of quality occurs considerably when restoring with only some bitstreams other than the bitstream. If the digital data is audio data, it will play an uncomfortable sound, and if it is video data, the image played on the screen will be distorted much.

The technical problem to be solved by the present invention to solve the above-described problems, the importance of the digital data and the additional information in order to be able to restore the data to the original data to minimize the degradation of the quality even if only a part of the bitstream The present invention provides a method and apparatus for encoding / decoding digital data which performs encoding / decoding in a high order.

1 is a block diagram showing the configuration of a digital encoding apparatus according to the present invention.

2 is a conceptual diagram illustrating a conventional digital encoding process.

3 is a conceptual diagram illustrating a bit division and encoding process according to the present invention.

4 is a block diagram showing the configuration of a decoding apparatus according to the present invention.

5 is a block diagram showing the configuration of an audio encoding apparatus according to the present invention.

FIG. 6 is a block diagram illustrating a detailed configuration of the bitpacking unit of FIG. 5.

7 is a block diagram showing the configuration of an audio decoding apparatus according to the present invention.

8 is a block diagram illustrating a configuration of a video encoding apparatus according to the present invention.

9A to 9E illustrate a process of processing a video signal in a video encoding apparatus.

10 is a block diagram showing the configuration of a video decoding apparatus according to the present invention.

In order to achieve the above technical problem, the digital data encoding method according to the present invention is a method of encoding a predetermined number of series of digital data, wherein the digital data is represented by a predetermined number of digits. Stage 1; And a second step of encoding by a predetermined encoding method in order from the most significant digit sequence consisting of the most significant digits of the digital data represented by the same number of digits to the less significant digit sequence.

The first step is to represent the digital data as binary data having the same number of bits, and the second step is to begin with a sequence of most significant bits (MSBs) having the highest importance of the represented binary data. An encoding step of encoding by a predetermined encoding method while descending to a lower lower bit sequence.

In the encoding step, the bits constituting the bit sequence are encoded in a predetermined number of bit units.

Another digital data encoding method according to the present invention is a method for encoding a predetermined number of digital data consisting of sign data and magnitude data, wherein each of the digital data is a predetermined number of digits. A conversion step represented by Digit; A first step of encoding a most significant digit sequence consisting of most significant digits of the highest importance digits of the size data constituting the expressed digital data by a predetermined encoding method; A second step of encoding code data corresponding to non-zero data among the encoded most significant digit sequences; A third step of encoding a digit sequence having the highest importance among unencoded size data of the digital data by a predetermined encoding method; A fourth step of encoding unsigned sign data among sign data corresponding to non-zero size data among the digit sequences encoded in the third step; And a fifth step of performing the third and fourth steps for each digit of the digital data.

In the converting step, the digital data is represented as binary data having the same number of bits, and the digits of the first to fifth steps are bits.

The encoding of the first to fifth steps is characterized by encoding the bits constituting each bit sequence of the size data and the sign data in a predetermined number of bit units.

In the digital data encoding apparatus according to the present invention for achieving the above technical problem, in a device for encoding a predetermined number of a series of digital data, each of the digital data is represented by a binary data consisting of a predetermined number of bits in units of bits A bit divider divided by; And an encoder which collects and encodes the bits having the highest importance from the bits divided by the bit splitter, and then collects and encodes the bits in the order of the bits having the highest importance. And a bit packing unit generating bit streams in order of importance of the data encoded by the encoder.

When the digital data is composed of sign data and magnitude data, the encoder collects and encodes size data of bits having the same rank as the same importance among the bit-divided data, and encodes the encoded size. It is preferable to encode unencoded code data among sign data corresponding to non-zero size data among the data, and to sequentially encode the size data and the code data from the most significant bit to the least significant bit.

The encoder is characterized in that the encoding by grouping the bits in a predetermined number of bits when collecting and encoding the bits according to the importance.

According to an aspect of the present invention, there is provided a digital data decoding method according to an embodiment of the present invention, wherein a method of decoding digital data encoded in the order of digits having the highest priority by assigning importance to digits of digital data includes: An analysis step of analyzing the importance; And a decoding step of decoding the analyzed digital data by a predetermined decoding method while descending from the high digit to the low digit.

The digit is characterized in that the bit.

The decoding step may include lossless decoding of the analyzed digital data in units of vectors while descending from bits of high importance to bits of low importance; And restoring data divided in bit units from the decoded data in vector units.

According to an aspect of the present invention, there is provided a digital data decoding method according to an embodiment of the present invention, wherein a method of decoding digital data encoded in the order of digits having the highest priority by assigning importance to digits of digital data includes: An analysis step of analyzing the importance; A first step of decoding the size data of the analyzed digital data from a digit of high importance to a digit of low importance by a predetermined decoding method; And a second step of decoding the sign data of the analyzed digital data and combining the same with the decoded size data.

The digit is characterized in that the bit.

The first step may include lossless decoding of the analyzed digital data in a vector unit while descending from a high importance bit to a low importance bit; And restoring data divided in bit units from the decoded data in vector units.

In the digital data decoding apparatus according to the present invention for achieving the above technical problem, in the device for decoding the digital data encoded in the order of the higher priority bit by attaching importance to the bits of the digital data, the encoded digital data bits A bitstream analyzer for analyzing the bit importance of the stream; A decoder which decodes the analyzed digital data from bits of high importance to bits of low importance; And a bit combiner for reconstructing digital data of each sample by combining information about each bit of each sample from the data decoded by the decoder.

The decoding unit may include: a size data decoding unit for decoding the size data of the analyzed digital data from a digit of high importance to a digit of low importance by a predetermined decoding method; And a sign data decoder which decodes the sign data of the analyzed digital data and combines the sign data with the decoded size data.

The decoder may losslessly decode the analyzed digital data from a bit of high importance to a bit of low importance, and restore data divided into bits.

According to another aspect of the present invention, there is provided an audio encoding apparatus, comprising: a time / frequency mapping unit configured to convert an input audio signal in a time domain into a frequency domain signal; A quantizer for quantizing the frequency domain signal for each predetermined frequency band; A bit divider dividing the quantized data into bit units; An encoder which collects and encodes the most significant bits (MSBs) among the data divided by the bit splitter, and subsequently collects and encodes the bits in the order of higher bits; And a bitstream generator for generating the encoded data in the encoder and additional information about the encoded data into a bitstream in order of importance of bits.

The bitstream generation sequence of the bitstream generator is characterized in that the generation of the bitstream in order of low frequency to high frequency.

According to another aspect of the present invention, there is provided an audio encoding method, comprising: dividing quantized audio data into bit units by a predetermined frequency band; Collecting and encoding the most significant bits (MSBs) of the bit-divided data, and subsequently collecting and encoding the bits in the order of higher bits; And generating the coded data and additional information about the coded data into a bitstream in order of importance of bits.

The bitstream generation sequence of the bitstream generation step is characterized in that the generation of the bitstream in the order of the low frequency to the high frequency, the high priority of the bit in the order of low to high.

In accordance with another aspect of the present invention, there is provided an apparatus for decoding audio data according to an embodiment of the present invention. The apparatus for decoding a bitstream of encoded audio data includes: a bitstream analyzer configured to analyze the importance of bits constituting the bitstream ; A decoder which decodes additional information and quantized data including at least quantization bits and sizes of quantization steps while descending from bits of high importance to bits of low importance according to the importance analyzed by the bitstream analyzer; An inverse quantizer for restoring the decoded quantization step size and the quantized data to a signal having an original size; And a frequency / time mapping unit for converting the dequantized signal into a signal in a time domain.

In accordance with another aspect of the present invention, there is provided a method for decoding audio data, by analyzing the importance of the bits constituting the bitstream and moving down from the higher priority bit to the lower priority bit. Decoding the side information including the size and the quantized data; Restoring the decoded quantization step size and quantized data to a signal having an original size; And converting the dequantized signal into a signal in a time domain.

According to another aspect of the present invention, there is provided a video data encoding apparatus comprising: a discrete cosine transform (DCT) unit for performing discrete cosine transform (DCT) on an input video signal; A quantizer for quantizing the DCT transformed data; And a bit packing unit configured to generate a bitstream by dividing the quantized data into additional information and information about quantized values according to importance.

According to another aspect of the present invention, there is provided a video data encoding method comprising: quantizing a discrete cosine transformed input video signal; Dividing the quantized data into additional information and information about quantized values in bit units according to importance; And generating the divided data according to the importance into a bitstream.

In accordance with another aspect of the present invention, there is provided a video data decoding apparatus comprising: a bitstream analyzer configured to analyze the importance of bits constituting the bitstream; A decoder which decodes additional information and quantized data including at least quantization bits and sizes of quantization steps while descending from bits of high importance to bits of low importance according to the importance analyzed by the bitstream analyzer; An inverse quantizer for restoring the decoded quantization step size and the quantized data to a signal having an original size; And an IDCT unit for inverse discrete cosine transforming the inverse quantized signal.

The decoding order of the decoder is characterized in that the order from the low frequency to the high frequency.

In accordance with another aspect of the present invention, there is provided a method of decoding video data, wherein in the method of decoding a bitstream of encoded video data, a high priority bit is analyzed by analyzing the importance of bits constituting the bitstream. Decoding the additional information and the quantized data including at least the quantization bit and the size of the quantization step while descending to bits of low importance; Inversely quantizing the decoded quantization step size and the quantized data to a signal having an original size; And inverse discrete cosine transforming the dequantized signal.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings. First, the overall concept of the encoding apparatus will be described. The input digital signal is made into a bitstream by the encoding apparatus shown in FIG. In this case, bits of high importance are encoded first to form a bitstream. That is, the encoding priority of data components to be encoded is determined according to the relative importance of the components. High priority components take precedence over low priority components. Since the information is encoded from such important information, if the number of bits used so far is equal to or larger than the allowed bit generation amount, the encoding may be stopped at that moment and the production of the bitstream may be finished. If the bitstream is interrupted in the middle, restoring the decoder causes some data to be lost, distorting the original digital data. However, even when the bitstream is finished in the middle as described above, since important information is encoded first, the overall performance can be maintained similarly to the conventional encoding method.

2 illustrates a general conventional encoding scheme. In the conventional method, the encoding is performed in order regardless of importance. Therefore, if it is necessary to use only a portion of the bitstream from the beginning of the entire bitstream, less significant information is included in the preceding bitstream than the information contained in the bitstream that cannot be used later.

For this reason, the present invention divides digital data into bit units as shown in FIG. In general, the significance of one bit of the most significant bit (MSB) is much higher than the importance of one bit of the least significant bit (LSB). Therefore, since the higher order bits of the quantized bits can be judged to be of higher importance, a method of encoding in order from the most significant bit to the least significant bit is used.

1 is a block diagram illustrating a configuration of an apparatus for encoding digital data according to the present invention, and includes a bit splitter 100, an encoder 110, and a bit packing unit 120. In this case, it is assumed that the digital data is composed of binary data and is represented by the same number of bits. If the digital data is not represented by the same number of bits, the digital data is first represented by the same number of bits. In the present embodiment, although the digital data is limited to binary data, it may be decimal or hexadecimal data other than binary data, and in this case, the digital data is expressed in digit units instead of bits.

The bit divider 100 divides a series of digital data into bit units. First, keep the sign value of digital data and take the absolute value of digital data so that the sign of all data becomes (+). When the absolute value data is represented as binary data, as shown in FIG. 3, values corresponding to each bit are divided separately according to the position of each bit, and data corresponding to each bit is collected separately to form a new sequence. For example, if the input digital data is -31, 12, -9, 7, 17, -23, ..., first taking the absolute value for each data, 31, 12, 9, 7, 17, 23, ... When the data (31, 12, 9, 7, 17, 23) whose absolute values are taken are represented in binary numbers, respectively, a, b, c, d, e, and f are represented.

31: 11111 _(a)

12: 01100 _(b)

9: 01001 _(c)

7: 00111 _(d)

17: 10001 _(e)

23: 10111 _(f)

In the values represented by the binary data, information about each bit is divided and collected in order to create a new sequence. First, when the data for the most significant bit is collected, the most significant bit of 31 is 1, the most significant bit of 12 is 0, the most significant bit of 9 is 0, ... and so on. Accordingly, the bit split data corresponding to the most significant bit becomes 1,0,0,0,1,1, .... The sequence of the next higher bit is 1,1,1,0,0,0, ... Subsequently, a sequence corresponding to each bit can be obtained. Finally, the sequence of least significant bits is 1,0,1,1,1,1, ...

The encoder 110 collects and encodes the bits having the highest importance, and encodes the binary data divided by the bit divider 100 by collecting the bits having the highest importance. The most significant bit is preferably the most significant bit (MSB) of each digital data represented as binary data, and the least significant bit is the least significant bit (LSB). The encoding is performed using a lossless encoding algorithm suitable for storing or transmitting the data.

In general, in order to compress more efficiently, data of the most significant bit are collected in order, several data are bundled and formed into a vector to form the most significant bits into vectors, and the vectors are used in a lossless encoding method. To encode it. As the lossless coding method, arithmetic coding or Huffman coding may be used. Thereafter, a bit-sliced coding scheme for collecting and encoding the next higher bits is used.

When the digital data is composed of sign data and magnitude data, the encoder 110 size data of the most significant bits (MSBs) among the data divided by the bit divider 100. Are encoded by encoding the sign data corresponding to non-zero size data among the encoded size data. This process is performed while descending to the least significant bit (LSB), but encoding sign data encodes only unsigned sign data.

In the present invention, since the absolute value of each sample value is first taken, information on whether a code value of each sample is (+) or (-) must be encoded first or later. In this case, the encoding of the code value before the encoding of less important information occurs first. The sign value is meaningless because the quantized value is regarded as 0 until the most significant bit is 1. That is, if the quantized value is represented by 5 bits of 00011 in binary and only the upper 3 bits of the 5 bits are used, the restored value of 00000 is restored. Therefore, this value will be useless information even if there is a sign bit. However, if 4 bits out of 5 bits are used, the value of 00010 becomes significant. The first 0 that comes out of the upper bits means that the quantized value is decoded into a value other than 0. Therefore, the sign value has a significant meaning.

When each frequency component is represented by the most significant bit, if the first non-zero 1 is displayed, the code value must be encoded according to whether the positive value or the negative value is followed before the other values.

For example, in the above example, when encoding the most significant bit, the first 1010 is encoded and then a determination is made as to whether the encoding of the code bit is necessary. In this case, since a non-zero value is first encoded in the first frequency component and the third frequency component, 0000 is encoded after the code bits of the two components are sequentially encoded.

The bit packing unit 120 makes the data encoded according to the importance in the encoder 110 into a bitstream having a desired size in the order of encoding. First, the encoded data of the bit splitting information for the most significant bit is made into a bitstream, and the information about the sign is encoded in the order as described above, and the information is put into the bitstream to produce the entire bitstream.

On the other hand, the bitstream generated through the encoding process is restored to the original digital data through the decoding process as shown in FIG. In the present invention, since the bitstream is formed by encoding the higher priority first, the decoder goes through a process of interpreting and decoding the bitstream in the order of high importance, that is, the order in which the bitstreams are produced.

4 is a block diagram illustrating a configuration of an apparatus for decoding digital data according to the present invention, and includes a bitstream analyzer 400, a decoder 410, and a bit combiner 420.

The bitstream analyzer 400 classifies the encoded data among the input bitstreams according to importance. Since the encoder makes the encoded data into the bitstream according to the importance of the data, the decoder analyzes the bitstream sequentially from the beginning according to the importance. First, the coded data for the bit splitting information for the most significant bit is classified in the bitstream, and the information for the sign is classified in the order as described above. The coded data classified as described above is transmitted to the decoder 410, and information about a sign is transmitted to the bit combiner 420, and used to recover the original signal.

The decoder 410 decodes the bit-divided signals from the signals from the bitstream analyzer 400 down to bits of high importance. The encoder may restore the bit-divided data from the encoded data through the inverse of the algorithm applied to encode the bit-divided data, that is, through the decoding process. In order to compress the data more efficiently, the encoder generally collects the data for the most significant bit in order, combines several data to form vectors, and uses the lossless coding method of the vectors. Therefore, these vectors are decoded from the coded data and the bit split data of each sample is reconstructed from the decoded vectors.

The bit combiner 420 restores the bit-divided binary data recovered by the lossless decoder 410 from the most significant bit information to the least significant bit information in order to make the original digital data. By filling data corresponding to each bit position of the digital data of each sample with data corresponding to each decoded bit, values obtained by taking absolute values of the original data can be obtained.

When the sign is (-) using the sign information of each sample obtained by the bitstream analyzer 400, the absolute value is multiplied by -1 so that the value becomes a (-) value.

For example, assume that the sequence for the decoded most significant bits is 1,0,0,0,1,1, ... If 5 bits are used to represent digital data, the digital data from the recovered bit division data is restored to 10000 ₍₂₎ , 00000 ₍₂₎ , 00000 ₍₂₎ , 10000 ₍₂₎ , 10000 ₍₂₎ , ... do. It then references the sequence for the next higher bit. If the sequence for the next higher bits is 1,1,1,0,0,0, ... the recovered digital data is 11000 ₍₂₎ , 01000 ₍₂₎ , 01000 ₍₂₎ , 10000 ₍₂₎ , 10000 ₍₂₎ , ... In this way, if the rest of the bit division data for the least significant bit is restored and the information for the sign is restored, the same data as the original input data can be restored.

5 is a block diagram illustrating the configuration of an audio encoding apparatus according to the present invention, and includes a time / frequency mapping unit 500, an acoustic psychology unit 510, a quantization unit 520, and a bit packing unit 530. .

The time / frequency mapping unit 500 converts an audio signal in a time domain into a signal in a frequency domain. The psychoacoustic unit 510 bundles the audio signals converted into signals in a frequency domain into signal components of an appropriate frequency band and calculates a masking threshold in each frequency band. The quantization unit 520 quantizes frequency signals of each frequency band so that the magnitude of quantization noise of each frequency band is smaller than the masking threshold calculated by the acoustic psychology unit 510. The bit packing unit 530 encodes the side information and the frequency signal component of the quantized frequency band to produce a bit stream.

The core of the present invention is a bit packing unit 530 for encoding a quantized data in the audio encoding device to create a bit stream. The encoding method of quantized digital data according to the present invention can be applied to the audio encoding apparatus shown in FIG.

Before quantizing the audio signal in the audio encoding apparatus, the acoustic psychology unit 510 first uses a psychoacoustic model to block the block type (long block, start block, short block, stop block) of the input data frame. Signal to Masked Thershold Ratio (SMR) value of each quantization band, area information in case of short block, and time delayed PCM data to synchronize the psychoacoustic model with time / frequency. Transfer to time / frequency mapping unit 500. The method for calculating the psychoacoustic model uses Model 2 of ISO / IEC 11172-3.

In addition, the time / frequency mapping unit 500 converts the data in the time domain into the data in the frequency domain using a modified DCT (Modified DCT) according to the block type output from the acoustic psychology unit 510. At this time, the size of block is 2048 for long / start / stop block and 8 times for MDCT with size 256 for short block. The procedure so far uses the same method used in the existing MPEG-2 NBC.

The quantization unit 520 bundles the frequency components into the quantization bands as shown in Table 1 by converting the data transformed into the frequency domain, so that the SNR (Signal-to-Noise Ratio) value of the quantization band is an SMR value that is an output value of the acoustic psychology unit 510. Quantize while increasing the step size to be smaller. Quantization uses scala quantization, and the default quantization step size interval is 21/4. Quantization is performed such that the NMR value is 0 dB or less. The output obtained at this time is information about the quantized data and the quantization step size of each processing band. In order to encode the quantized signals, first, the largest quantization bit required for encoding is calculated by finding the largest absolute value of the quantized signals for each coding band.

FIG. 6 is a block diagram illustrating a detailed configuration of a bitpacking unit 530 in which a quantized data encoding method of the present invention is applied, not a conventional method. The bit splitter 600 and the encoder are shown in FIG. 610 and the bitstream generator 620.

The bit splitter 600 divides the quantized data output from the quantizer 520 in units of bits. The encoder 610 collects and encodes the most significant bits (MSBs) among the bit-divided data output from the bit divider 600, and then encodes the bits in the order of higher bits. The bitstream generator 620 generates the coded data output from the encoder 610 and additional information about the encoded data in a bitstream in order of importance of bits.

Meanwhile, the encoding of the additional information and the quantized data in the bit splitter 600 and the encoder 610 will be described in detail as follows. The synchronization signal of the bitstream is put into the bitstream to generate information about the start of the bitstream. Then, first, the size of the entire bitstream is encoded. Next, the block type for which type of block is used must be encoded. The subsequent encoding process differs slightly depending on the shape of the block. According to the characteristics of the signal, it is divided into a case of transforming a large block to encode an input signal of one frame and a case of converting into 8 short blocks. As the size of the block changes, there is a slight difference in the encoding process.

First, the maximum quantization bit value is obtained from the quantized signals of each encoding band as shown in Table 1, and the encoding is performed by the bit division encoding method, which is the method proposed by the present invention, from the maximum quantization bit value. The quantization step size information of the angularization band generated while quantizing is encoded. In the method of encoding quantization bits or quantization step size information, first, a minimum value and a maximum value of the quantization bit or quantization step sizes are obtained, and the size of the difference between the two values can be obtained to determine the number of bits required. Before encoding this additional information, the minimum value and the size required for bit representation are first encoded by arithmetic encoding and stored in the bitstream. When the actual encoding is performed later, the difference between the additional information and the minimum value is encoded. . Then, the next quantized signal is encoded.

The coding method of the quantized signal uses a bit-sliced coding method in which each quantized signal is grouped in units of bits and then a vector is formed by grouping bits having the same importance and encoding the vector. When sign information of each sample is needed in the middle of encoding the quantized signals, sign information is also encoded. When encoding is performed according to the importance, if the number of bits used so far is equal to or larger than the allowed bit amount, encoding is stopped at that moment and the bitstream is completed. Therefore, the overall complexity can be significantly lower than that of the conventional coding scheme. However, if you want to perform the encoding with better performance regardless of the complexity, increase the size of the bit amount after encoding by increasing the step size of each quantization band appropriately when the used bit amount is larger than the allowed bit amount. It can be reduced as appropriate. The overall performance can be improved by repeating the quantization and encoding processes until the generated bit amount is smaller than the allowed bit amount.

Similarly, in the case of a short block, it is divided into eight blocks, which are 1/8 of a long block, and subjected to time / frequency mapping and quantization, and then lossless coding the quantized data. The quantization process is not performed for every eight short blocks, but using information obtained by dividing the eight blocks sent from the psychoacoustic unit into three regions, and collecting the quantization bands as shown in Table 2 in this region, Band fringe treatment. Thus, quantization step size information for each band of three regions is obtained. First, quantization bit information is encoded. The maximum quantization bit can be obtained and the bit division coding of the present invention can be performed like the long block. Of course, if a quantization bit of a band is smaller than the quantization bit currently being coded, no encoding is passed and the quantization band is encoded only when the quantization bit is equal to the quantization bit of the band. In this case, in the case of encoding a band for the first time, stepsize information on the quantization band is first encoded, and then values corresponding to the quantization bits are extracted from the values of the quantized frequency components.

Through such a series of processes, the encoding is performed in the order of high importance as a whole, and the bitstream generator 620 configures the bitstream. That is, the bitstream is encoded in order of synchronization information (sync), frame size, block type, quantization bits of each coding band, quantization step size information of each quantization band, and least significant bits in the most significant bit signals of the quantized audio signal. Will be configured.

FIG. 7 is a block diagram illustrating a configuration of an audio decoding apparatus for decoding a bitstream generated by the audio encoding apparatus. The bitstream analyzer 700, the decoder 710, the inverse quantizer 720, and the frequency time are shown in FIG. It consists of a mapping unit 730.

The decoding order of the audio bitstream in the audio decoding apparatus is the reverse of the encoding process of the audio encoding apparatus. The bitstream analyzer 700 analyzes the importance of the bits constituting the bitstream. The decoder 710 is a block for decoding the quantization bits, the size of the quantization step, the quantized data, etc. according to the order in which the bitstreams input from the audio encoding apparatus are described above, and the bitstream analysis unit 700 analyzes them. The additional information including the size of the quantization bit and the quantization step and the quantized data are decoded while descending from the high importance bit to the low importance bit according to the importance level. The dequantization unit 720 restores the decoded quantization step size and the quantized data to a signal having an original size. The frequency / time mapping unit 730 converts the dequantized signal into a signal in the time domain so that the user can reproduce the signal.

8 illustrates a block diagram of a video encoding apparatus, and includes a DCT unit 800, a quantization unit 810, and a bitpacking unit 820. A feature of the present invention resides in a bit packing unit 820 that generates a bitstream by encoding quantized data in the video encoder. The encoding method of quantized digital data according to the present invention can be applied to the audio encoder shown in FIG.

The DCT unit 800 performs discrete cosine transform (DCT) on a video signal in a spatial domain having an arbitrary size and converts the video signal into data in a frequency domain. The quantization unit 810 quantizes the transformed frequency domain data. The bitpacking unit 820 divides side information and information about quantized values into bit units according to the importance of video data to be encoded, and encodes the bitstream in the order of the most significant bit and then from a low frequency to a high frequency. To create.

That is, as shown in FIG. 8, data of a frequency domain is obtained from a video signal of a spatial domain having an arbitrary size using a DCT transform, and then quantized in an appropriate amount through a quantizer, and the quantized data is shown in FIG. You can do the array as shown.

As shown in FIG. 9B, the data of the spatial region of MxN having the data structure as shown in FIG. 9A is divided into arbitrary blocks of PxQ (generally 16x16). The PxQ block is then divided into four uniformly sized subblocks (generally 8x8). Then, when DCT is performed through the DCT unit 800 and converted into the frequency domain, coefficients of frequency components having 64 bands can be obtained as shown in FIG. When this is called F _i (i = 0,1,2,..., 63), it is quantized through the quantization unit 810. If the quantized data is P (F _i ) (where i = 0, ..., 63), these values are arranged on a one-dimensional array, as shown in FIG. Generally, the low frequency component exists in most blocks, but the high frequency component exists only in a few few blocks. Thus, the size of data has a shape as shown in FIG.

After arranging the quantized data in this manner, encoding may be applied to the rearranged quantized data by applying the digital encoding method according to the present invention through the bit packing unit 820. Then, the bitstream can be produced by efficiently encoding the original quantized data according to the importance.

FIG. 10 is a block diagram illustrating a configuration of a video decoding apparatus for decoding a bitstream encoded by the video encoding apparatus. The bitstream analyzer 10, the decoder 20, the inverse quantizer 30, and the IDCT unit are illustrated in FIG. It consists of 40.

The bitstream analyzer 10 analyzes the importance of bits constituting the encoded bitstream. The decoder 20 adds additional information and quantized data including at least the quantization bit and the size of the quantization step while descending from a high importance bit to a low importance bit according to the importance analyzed by the bitstream analyzer 10. Decrypt That is, the bitstream produced by the video encoder is input, and the decoder first decodes the bit-divided data according to the importance in the manner proposed in the present invention, and decodes each bit-divided quantized data according to the importance and quantizes each frequency component with the decoded data. The combined data can be combined. By rearranging the combined quantized data of the 64 frequency components in the reverse order rearranged by the encoder, the quantized data of the frequency components of the original subblock can be converted. The inverse quantization unit 30 restores the decoded quantization step size and the quantized data to a signal having an original size. The IDCT unit 40 reconstructs the inverse quantized signal by inverse discrete cosine transform (IDCT) to a video signal of a spatial domain.

According to the present invention, deterioration of sound quality can be reduced even if a part of the bitstream is lost or corrupted by encoding important information first.

In addition, since the present invention is a new method of lossless encoding of digital data, data encoded by another lossless encoding method may be converted into the method of the present invention, and thus may be compatible with existing encoding methods.

In addition, since the more important information is encoded first, the present invention can be applied to encoding various types of signals such as encoding of an audio signal and encoding of a video signal.

Claims (19)

In the method of encoding a series of digital data consisting of sign data and magnitude data, A first step of representing each of the digital data as binary data having the same number of bits; A second step of encoding, by a predetermined encoding method, a most significant bit sequence consisting of most significant bits of size data constituting the expressed digital data; A third step of encoding code data corresponding to non-zero data among the encoded most significant bit sequences; A fourth step of encoding a bit sequence having the highest importance among unencoded size data by a predetermined encoding method; A fifth step of encoding uncoded code data among the bit sequences encoded in the fourth step; And And a sixth step of performing the fourth and fifth steps on each bit of the digital data. The method of claim 1, wherein the encoding of the second to sixth steps is performed. And encoding the bits constituting the bit sequence for the size data and the sign data by grouping them in a predetermined number of bit units. The method of claim 2, wherein the predetermined encoding method Digital data encoding method characterized in that the lossless coding. 4. The method of claim 3, wherein the lossless coding Digital data coding method characterized in that the Huffman coding (Huffman coding). 4. The method of claim 3, wherein the lossless coding Digital data coding method characterized in that the arithmetic coding (Arithmetic coding). In the device for encoding a series of digital data consisting of sign data and magnitude data, A bit divider for dividing each of the digital data into binary data consisting of a predetermined number of bits and dividing the digital data into bit units; The bits having the highest importance are collected and encoded from the bits divided by the bit splitter, and the bits are collected and encoded in the order of the higher priority bits, and the bits having the same priority among the divided bits are assigned to bits having the same priority. An encoding unit which collects and encodes the size data for the encoded data and encodes corresponding uncoded code data; And And a bit packing unit for generating a bitstream in order of importance of the data encoded by the encoder. The method of claim 6, wherein the encoder And collecting and encoding the bits according to importance, by encoding them in a predetermined number of bit units. An apparatus for decoding digital data encoded in the order of the higher priority bits by attaching importance to bits of digital data, A bitstream analyzer for analyzing bit importance of the encoded digital data bitstream; A decoder which decodes the analyzed digital data from bits of high importance to bits of low importance; And And a bit combiner for reconstructing digital data of each sample by combining additional information of each bit of each sample from the data decoded by the decoder. The method of claim 8, wherein the decoding unit A size data decoding unit for decoding the size data of the analyzed digital data from a digit of high importance to a digit of low importance by a predetermined decoding method; And And a sign data decoder which decodes the sign data of the analyzed digital data and combines the sign data with the decoded size data. The method of claim 8 or 9, wherein the decoding unit And digitally recovering the data divided by the bit unit by lossless decoding the analyzed digital data from the high bit to the low bit. A time / frequency mapping unit converting an input audio signal in a time domain into a frequency domain signal; A quantizer for quantizing the frequency domain signal for each predetermined frequency band; A bit divider dividing the quantized data into bit units; An encoder which collects and encodes the most significant bits (MSBs) among the data divided by the bit splitter, and subsequently collects and encodes the bits in the order of higher bits; And And a bitstream generator configured to generate a bitstream in order of importance of bits from the encoded data and additional information about the encoded data. 12. The apparatus of claim 11, wherein the bitstream generator An audio encoding apparatus characterized by generating a bitstream in the order of a low frequency to a high frequency. Dividing the quantized audio data into bit units by a predetermined frequency band; Collecting and encoding the most significant bits (MSBs) of the bit-divided data, and subsequently collecting and encoding the bits in order from the higher bits; And And generating the encoded data and additional information about the encoded data into a bitstream in order of importance of bits. The method of claim 13, wherein the bitstream generation order of the bitstream generation step An audio encoding method comprising generating bits in order of high to low frequency in order of low frequency to high frequency. An apparatus for decoding a bitstream of encoded audio data, A bitstream analyzer for analyzing the importance of bits constituting the bitstream; A decoder which decodes additional information and quantized data including at least quantization bits and sizes of quantization steps while descending from bits of high importance to bits of low importance according to the importance analyzed by the bitstream analyzer; An inverse quantizer for restoring the decoded quantization step size and the quantized data to a signal having an original size; And And a frequency / time mapping unit for converting the dequantized signal into a signal in a time domain. In the method for decoding a bitstream of the encoded audio data, Analyzing the importance of bits constituting the bitstream; Decoding the additional information and the quantized data including at least the quantization bit and the size of the quantization step while descending from the high importance bit to the low importance bit; Restoring the decoded quantization step size and quantized data to a signal having an original size; And And converting the dequantized signal into a signal in a time domain. An apparatus for decoding a bitstream of encoded video data, A bitstream analyzer for analyzing the importance of bits constituting the bitstream; A decoder which decodes additional information and quantized data including at least quantization bits and sizes of quantization steps while descending from bits of high importance to bits of low importance according to the importance analyzed by the bitstream analyzer; An inverse quantizer for restoring the decoded quantization step size and the quantized data to a signal having an original size; And And an IDCT unit for inverse discrete cosine transforming the dequantized signal. The method of claim 17, wherein the decoding unit A video decoding apparatus characterized by decoding from low frequency to high frequency. In the method for decoding a bitstream of the encoded video data, Analyzing the importance of the bits constituting the bitstream to decode the additional information and the quantized data including at least the quantization bit and the size of the quantization step while descending from the higher priority bit to the lower priority bit; Inversely quantizing the decoded quantization step size and quantized data to restore a signal having an original size; And And inverse discrete cosine transforming the inversely quantized signal.

Images