WO2012159301A1

WO2012159301A1 - Method, device and medium for video entropy encoding and entropy decoding

Info

Publication number: WO2012159301A1
Application number: PCT/CN2011/076455
Authority: WO
Inventors: 徐苏珊
Original assignee: 深圳市融创天下科技股份有限公司
Priority date: 2011-05-25
Filing date: 2011-06-28
Publication date: 2012-11-29
Also published as: CN102238387B; CN102238387A

Abstract

The present invention discloses a method, device and medium for video entropy encoding and entropy decoding. The entropy encoding method includes the following steps of: acquiring a syntactic element to be encoded wherein the syntactic element is in a block to be encoded; converting the syntactic element to be encoded into a code index; converting the code index into a codeword number; and encoding the codeword number according to a codeword code rule and acquiring a final codeword. By introducing code indexes and regular variable length/fixed length encoding methods, the present invention newly designs each syntactic element code table and realizes quick code table search by the code indexes, thereby avoiding the problems of that the conventional variable length code table requires to occupy biggish memory space and requires a bitwise traverse operation. On the premise of ensuring encoding efficiency, the present invention greatly reduces implementation complexities of entropy encoding and entropy decoding operations on a mobile terminal apparatus. The present invention is applied to a portable terminal whose hardware resources are restricted.

Description

Video entropy coding, entropy decoding method, device and medium

Technical field

The present invention relates to the field of video coding technologies, and in particular, to a video entropy coding, an entropy decoding method, an apparatus, and a medium.

Background technique

International standards for video compression, such as MPEG-x and H.26x, are block-based hybrid compression algorithms that eliminate video temporal redundancy by motion estimation/motion compensation, discrete cosine transform transformation of difference images to eliminate spatial redundancy, and quantization The coefficients after entropy coding eliminate statistical redundancy. Entropy codec is an important part of video coding. Entropy coding is a series of syntax elements generated by various functional modules in video coding algorithms. Its purpose is to effectively remove statistical redundancy of data. These syntax elements can be classified into two categories: residual coefficient syntax elements and non-residual coefficient syntax elements. The residual coefficient syntax element has the largest amount of information and is the processing focus of entropy coding. The syntax elements of the residual coefficient are obtained by scanning the block residual coefficients after transformation and quantization. The syntax elements of the non-residual system are generated by video coding links such as intra prediction, inter prediction, and quantization, and input to the entropy coding module for processing. The syntax elements of the non-residual system include macroblock type, reference frame index, and quantization. Parameters, motion vector residuals, coding models, intra prediction modes, etc. In video coding, a macroblock is the basic unit of video coding. The macroblock size is 16 X 16, each macroblock consists of four 4 X 4 blocks, each block consists of one 4 X 4 luma block and two colors. The block is composed.

The encoding methods for block prediction residual coefficients of MPEG-1, H.261, MPEG-2, H.263, MPEG-4, and the latest H.264 video standards are classified as follows:

(1) 2D-VLC (Variable Length Coding) entropy coding of H. 261, MPEG-1, and MPEG-2 standards;

(2) H. 263, 3D-VLC entropy coding of the MPEG-4 standard;

(3) H. 264/AVC standard context-based adaptive variable length coding (CAVL) and context-based adaptive binary arithmetic coding CABAC (Context-Based Adaptive Binary Arithmetic Coding, CABAC) .

After the residual coefficient of the image block is processed by the transform quantization step, a two-dimensional residual coefficient is obtained. For further processing, the two-dimensional coefficient is generally converted into a one-dimensional sequence by scanning; There are many ways to represent a dimensional sequence. Different representation methods define their own syntax elements. Conversely, a one-dimensional sequence can be reconstructed using syntax elements. The residual coding parameters associated with these syntax elements are: Level, Run, and End of Block EOB (End of Block). Usually, the entropy coding first performs Zigzag scanning on the quantized coefficients of the prediction transform, and records the non-zero coefficients encountered in the scan as Level, and the continuous zero-coefficient run before the non-zero coefficients is recorded as Run.

Specifically: 2D-VLC entropy coding uses joint coding (Level, Run) data pairs of residual coefficients, and E0B coding, E0B is assigned a codeword separately; 3D-VLC entropy encoder integrates data end information into (Level , Run) data alignment, form a three-dimensional data set (Level, Run, Last), Last indicates whether the current coefficient is the last coefficient in the block, so that you can not encode the end identifier, and play the combination of Level, Run and E0B. Probability; The CAVLC entropy coding of the H.264/AVC video coding standard makes full use of the context information. The residual information of 4×4 blocks is optimized by establishing multiple code tables, and the Level and Run are separately coded. A good coding effect has been achieved. The above entropy coding methods are all variable length coder, and the basic principle of coding still follows the Huffman coding principle, that is, a large probability symbol assigns a short codeword, and a small probability symbol assigns a long codeword, thereby achieving an average code. The shortest purpose. H. 264/AVC CABAC entropy coding belongs to arithmetic coder, the performance of arithmetic coding is better than variable length coding, and CABAC is closer to Shannon's theorem through adaptive probability estimation and update and binary arithmetic coding, and the coding effect is higher. However, the computational complexity of the algorithm is high, and the hardware implementation requirements are complex. It is an improvement of the coding efficiency by the complexity of software and hardware.

It can be seen that the above video coding standard can obtain higher coding efficiency by performing context-based coding by analyzing the probability distribution of the block prediction residual coefficients in different states. The CAVLC in the H.264/AVC standard makes full use of the probability and statistical distribution of different residual coefficients, and adaptively uses a plurality of code tables to encode a corresponding syntax element. CAVLC is the most efficient coding in the currently published video standard. Context adaptive variable length coding techniques. For applications such as real-time video communication such as mobile video, the entropy codec technology uses the exponential Columbus code Exp-Golomb and CAVLC, and the index Columbus code mainly encodes non-residual information such as motion vectors; CAVLC coding is mainly for prediction. The residual is encoded. Both the index Columbus code and the CAVLC code conform to the basic idea of Huffman coding: short code words represent high probability information and long code words represent low probability information. However, the index Columbus coding is a regular variable length coding based on symbolic probability statistics. Based on the traditional variable length coding method, CAVLC defines multiple sets of different code tables by introducing content adaptive models. Word features, dynamic code table switching, improve coding efficiency. The decoding process of CAVLC is too large. On the one hand, CAVLC decoding must distinguish codewords of different lengths from successive bitstreams. For example, the full search algorithm needs to traverse the entire binary tree, which requires multiple readings and judgments. Table process consumption On the other hand, because the codeword tree is sparse, the storage space of the codebook grows exponentially with the increase of the codeword length, but most of them are vacant and are not effectively utilized.

The entropy coding of H.264 achieves good coding efficiency through good context adaptive design, but these are at the cost of too many code tables. Too many code tables mean that the codec operation requires more computing resources. . Especially for the mobile video service system, due to factors such as the computing hardware of the portable mobile terminal and the limited resources of the battery power consumption, the high complexity video entropy decoding implementation still faces many problems: many code tables of the entropy decoder need to be searched. The LUT (Look Up Table) structure is implemented, which occupies a large memory space. The entropy decoding process often needs to traverse the lookup table to match the codewords bit by bit. Repeated read and write memory operations will also lead to a sharp increase in power consumption.

Summary of the invention

The purpose of the embodiment of the present invention is to provide a video entropy coding method, which aims to solve the problem that the conventional variable length code table occupies a large memory space, requires a bit-by-bit traversal operation, and has high complexity of entropy coding and decoding.

The method of the embodiment of the present invention is implemented as a video entropy coding method, where the method includes: acquiring a syntax element to be encoded in a block to be coded;

Converting the syntax element to be encoded into a coding index;

Converting the encoded index into a codeword value;

The codeword value is encoded according to a codeword encoding rule to obtain a final codeword.

A second object of the embodiments of the present invention is to provide a video entropy coding apparatus, where the apparatus includes: an encoding syntax element acquiring module, configured to acquire a syntax element to be encoded in a block to be encoded; Decoding the coding syntax element into a coding index;

a second conversion module, configured to convert the encoding index into a codeword value; the mapping manner of converting the encoding index into a codeword value comprises a probability statistical distribution mapping or a non-probability statistical distribution mapping; And a module, configured to encode the codeword value according to a codeword encoding rule to obtain a final codeword.

A third object of embodiments of the present invention is to provide a recording medium including the program of the video entropy encoding method.

A fourth object of the embodiments of the present invention is to provide a video entropy decoding method, where the method includes: reading an entropy encoded binary bit stream, performing calculation according to a codeword encoding rule, and obtaining a complete Binary code word

Calculating the codeword value according to the codeword encoding rule by using the binary codeword;

Converting the codeword value into an encoding index;

Corresponding syntax elements are obtained according to the encoding.

A fifth object of the embodiments of the present invention is to provide a video entropy decoding apparatus, where the apparatus includes: a complete codeword acquisition module, configured to read an entropy encoded binary bit stream, perform calculation according to a codeword encoding rule, and obtain a complete Binary code word;

a codeword value obtaining module, configured to calculate, according to the codeword encoding rule, the codeword value;

a third conversion module, configured to convert the codeword value into an encoding index;

The second syntax element obtaining module is configured to obtain a corresponding syntax element according to the encoding index.

A sixth object of the embodiments of the present invention is to provide a recording medium including the program of the video entropy decoding method.

Advantageous effects of the present invention

In the embodiment of the present invention, the syntax element to be encoded in the block to be encoded is obtained, and the syntax element to be encoded is converted into a coding index, and the coded index is converted into a codeword value, and the codeword value is regularly lengthened. Coding or fixed length coding, obtaining the final codeword; The key idea of the embodiment of the present invention is to replace the context adaptation probability distribution of the syntax element by the regular codeword, and replace by introducing the coding index

CAVLC's many code table designs eliminate the need for the traditional variable length code table to occupy a large memory space and need to be traversed bit by bit. Under the premise of ensuring coding efficiency, the entropy coding and entropy decoding on the mobile terminal device are greatly reduced. Implementation complexity, suitable for portable terminals with limited hardware resources.

DRAWINGS

1 is a flow chart of a preferred embodiment of a video entropy encoding method of the present invention;

2 is a Zigzag scanning sequence diagram of a prediction residual block of 4×4 luminance of a preferred embodiment of the video entropy encoding method of the present invention;

Figure 3 is a diagram showing the structure of a preferred embodiment of the video entropy coding apparatus of the present invention ^

4 is a flow chart of a preferred embodiment of a video entropy decoding method of the present invention;

Figure 5 is a diagram showing the structure of a preferred embodiment of the video entropy encoding apparatus of the present invention ^ The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. For the convenience of description, only the parts related to the embodiments of the present invention are shown. It is understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

In the embodiment of the present invention, the syntax element to be encoded in the block to be encoded is obtained, and the syntax element to be encoded is converted into a coding index, and the coded index is converted into a codeword value, and the codeword value is regularly lengthened. Encoding or fixed length coding, obtaining the final codeword; In the embodiment of the present invention, by introducing a coding index and a regular variable length/fixed length coding method, a code table of each syntax element is newly designed, and a fast code table retrieval is realized by the coding index to eliminate The traditional variable length code table needs to occupy a large memory space and needs to be traversed bit by bit. Under the premise of ensuring coding efficiency, the implementation complexity of the entropy decoding operation on the mobile terminal device is greatly improved, and the hardware resource is limited. Portable terminal. Embodiment 1

FIG. 1 is a flowchart of a video entropy coding method according to an embodiment of the present invention, where the method includes the following steps:

S101. Obtain a syntax element to be encoded in the block to be encoded.

The syntax element to be encoded includes: a residual coefficient syntax element or a non-residual coefficient syntax element; the method for obtaining the residual coefficient syntax element is as follows:

A) reordering the residual coefficients of the block to be coded after prediction and quantization according to the scanning order, and arranging the two-dimensional residual coefficients into a one-dimensional sequence;

The scanning sequence is a Zigzag scan reordering method;

B) acquiring a block residual coefficient syntax element to be coded according to the one-dimensional sequence;

Obtaining, according to the one-dimensional sequence, a block residual coefficient syntax element is: obtaining a corresponding syntax element according to a statistical feature of the residual coefficient in the one-dimensional sequence; and the corresponding syntax element can represent the one-dimensional sequence.

A grammar element can represent all the information of a one-dimensional sequence, that is, it only reconstructs a one-dimensional order by relying on grammatical elements. Column. However, as part of the entropy coding of video coding, it is often desirable to use the statistical characteristics of the residual coefficients to reduce the number of bits in which the information is stored. The following example illustrates:

Example: A 16-element one-dimensional sequence data string: 0, 3, 0, 1, -1, -1, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0. The idea of using run-length coding: Let 0 run as run, non-zero coefficient be level, and use (run, level) run pairs to represent the one-bit sequence is (1, 3), (1, 1), (0, - 1), (0, -1), (1, 1) and a terminator E0B. It can be seen that to preserve 16 one-dimensional sequence residual information losslessly, the required number of bits of information is 16xH (H is the number of binary digits required for each coefficient preservation); and after run-length coding, 5 (run, Level ) Run pair, if the run pair is defined as a syntax element, regardless of the probability distribution of the run-pair/syntax element (the equal probability case, the entropy value is the largest), assign H bits to each data in the run-pair pair, required The total information bit is HxH, where the terminator E0B is assigned the H bit). Therefore, the number of bits required is much less than the number of bits required to save the original sequence. Moreover, the probability distribution of the grammatical elements is actually a non-probability distribution whose entropy value is less than the entropy value of the equal probability, in other words, all data can be saved with fewer bits.

For example, the syntax element acquisition manner of the one-dimensional sequence according to the H.264, MPEG_2, and MPEG-4 coding standards may be classified into a single data mode, a data pair mode, or a data group mode;

The residual coefficient syntax elements obtained by the single data method include: the number of non-zero coefficients TotalCoeffs, the number of trailing coefficients Trai l ingOnes, the number of zeros before the zero coefficient Total_zeros, the non-zero coefficient magnitude Level and run Run — before;

The residual coefficient syntax elements obtained by data pairing include: (Level, Run) data pair and data terminator;

The residual coefficient syntax elements obtained by the data group method include: a three-dimensional data group (Level , Run, Last );

The non-residual coefficient syntax element includes a macroblock type, a reference frame index, a quantization parameter, a motion vector residual, an encoding model, or an intra prediction mode;

S102, converting the syntax element to be encoded into an encoding index Codelndex;

The syntax element to be encoded includes: a residual coefficient syntax element or a non-residual coefficient syntax element; the method for converting the syntax element to be encoded into a coding index may be a probability statistical distribution mapping method or a non-probabilistic statistical distribution mapping method; A probability and statistical distribution mapping method for converting a syntax element to be encoded into a coding index, specifically:

A) obtaining probability statistical distribution information of the syntax element to be encoded in the block to be coded; the probability distribution information is a probability distribution statistical value of the syntax element to be encoded;

Specifically:

1) obtaining a prediction residual quantized block residual coefficient;

2) counting a probability statistical distribution value of the syntax element based on the context model according to the residual coefficient of the block to be encoded, as the probability statistical distribution information of the syntax element to be encoded;

The context model is a joint probability established based on the interdependence of information before and after. The front and rear information indicates the processed data unit and the current data unit to be processed, including the image group, the frame, the slice, the block, and the like. The front and rear information in the embodiment of the present invention is the processed block and the current to-be-processed block; Interdependencies include: a) Spatial correlation between neighboring blocks; Since neighboring blocks in the same frame tend to have a large degree of spatial correlation, information of the current block can be predicted by neighboring blocks. b) The mathematical constraints inherent in the grammatical elements. In video coding, the context model is generally referred to as context. The theory of information theory is that the joint entropy is smaller than the sum of the source entropy, which further improves the compression efficiency.

Further, the probability statistical distribution information of the probabilistic statistical distribution values may be combined to obtain a probability statistical distribution value of the merged syntax elements as the probability statistical distribution information of the syntax elements to be encoded.

B), performing probability ordering on the syntax elements to be encoded according to the probability statistical distribution information; the probability ordering may be ascending or descending;

0, establishing a mapping relationship between the syntax element to be encoded and an encoding index;

In the step of “establishing a probability and statistical distribution map of syntax elements and coding indexes”, establishing a mapping of non-residual coefficient syntax elements and coding indexes may be mapping of a single non-residual coefficient syntax element and a coding index Codelndex, or Is to create multiple non-residual coefficient syntax elements and coding index

Joint mapping of the mapping of Codelndex.

The "joining a joint map of a plurality of non-residual coefficient syntax elements and a code index of a code index" may be used to synthesize a joint probability distribution of a plurality of non-residual coefficient syntax elements to establish a corresponding A non-residual coefficient syntax element mapping table, or grouping the non-residual coefficient syntax elements and performing comprehensive coding.

further,

A non-probabilistic statistical distribution mapping method that converts non-residual coefficient syntax elements in a syntax element to be encoded into a coding index includes unsigned direct mapping, signed conversion to symbol mapping, or mapping table mode:

(1) Unsigned direct mapping. The syntax elements of this mapping method include macroblock type, parameter frame index, and the like.

CodeNum=CodeIndex=v, v is the value of the non-residual coefficient syntax element

(2) Signed into a symbol map. The syntax elements of this mapping method mainly include motion vector difference values, quantization factor increments, and the like.

(3) Mapping table. The mapping relationship between the non-residual coefficient syntax element and the second coding index Codelndex is specified by formulating a mapping table. It is mainly used for mapping of parameters such as block coding mode and intra prediction mode.

Table 7 is a mapping table relationship table of the X 4 luma block intra prediction mode and the encoding index Codelndex according to the embodiment of the present invention. The table maps a single non-residual coefficient syntax element and an encoding index.

Table 7

S103: Convert the code index Codelndex into a codeword value CodeNum; the mapping method for converting the code index Codelndex into a codeword value CodeNum may be a probability statistical distribution mapping method or a non-probabilistic statistical distribution mapping method;

The probability distribution distribution mapping method of converting the coding index Codelndex into the codeword value CodeNum is: establishing a mapping relationship between the coding index and the codeword value according to the probability statistical distribution information of the syntax element to be encoded; wherein the mapping relationship follows The principle of the shortest average code length, the codeword value and the codeword of the mapped by the probability of the syntax element to be encoded: the short codeword is assigned to the large probability codeword value, and the long codeword is assigned by the small probability codeword value;

The non-probabilistic statistical distribution mapping method for converting the encoding index Codelndex into a codeword value CodeNum is:

The code index Codelndex is linearly associated with the codeword value CodeNum, that is, CodeIndex=axCodeNum+p (where a and β are integers); or the mapping relationship between the coded bow I and the codeword value is established by formulating a mapping table;

S104: Encoding the codeword value CodeNum according to a codeword encoding rule to obtain a final code. The codeword encoding rule may be a regular variable length coding or a regular fixed length coding;

The codeword has a regular codeword structure, and the codeword value CodeNum and the codeword CodeWord directly have a determined mathematical relationship, and the codeword can be generated by calculating the codeword value, and the codeword value can also be calculated by the codeword;

The regular variable length coding may be a Columbus-Golomb-Rice or an index Columbus coded Exp-Golomb coding rule of different order; Golomb-Rice and Exp-Golomb are both Columbus codes, and the order is a Columbus coded parameter. , different orders have different coding formulas.

The codeword is in a binary format, including a prefix and a suffix portion, and the prefix generally begins with a plurality of 0s plus a 1;

The embodiment of the present invention lists the partial coding structure of the Columbus-Golomb-Rice and the index Columbus code Exp-Golomb. As shown in Table 1, the partial codewords of Golomb-Rice are shown as the order k=0, 1, 2, 3. Table 2 shows the partial codewords of the Exp-Golomb code when the order k=0, 1, 2, 3. Table 1

Table 2

In the embodiment of the present invention, by introducing a coding index and a regular variable length/fixed length coding method, each syntax element code table is newly designed, and a fast code table retrieval is realized by the coding index, and the rule variable length/fixed length coding does not need to save the actual binary. The codeword eliminates the need for a large memory space and requires a bit-by-bit traversal operation. Under the premise of ensuring coding efficiency, the implementation complexity of the entropy decoding operation on the mobile terminal device is greatly improved, and is applicable to A portable terminal with limited hardware resources. The following is a detailed description of the implementation process of the present invention by taking an entropy coding of a 4×4 luma prediction residual block using H.264/AVC CAVLC entropy coding. 2 is a Zigzag scanning sequence diagram of a prediction residual block of the luminance of the embodiment 4X4 according to the embodiment of the present invention, wherein the numerical values in the squares indicate the luminance residual values, and the arrows indicate the scanning order.

The residual coefficients of the 4X4 block are mapped to a one-dimensional sequence of 16 elements by Zigzag order: 0, 3, 0, 1, -1, -1, 0, 1, 0, 0, 0, 0, 0, 0, 0 , 0.

If the H.264/AVC CAVLC entropy coding is used, the residual coefficient syntax elements that need to be encoded include:

(a) The number of non-zero coefficients (TotalCoeffs) = 5 and

The number of trailing coefficients (Trailing0nes) = 3;

(b) The number of zeros before the non-zero coefficient (Total_zeros) = 3;

(c) Level (i) = l i = l, 0;

(d) Run— before (i) i=4, 3, 2, 1, 0.

Following the CAVLC coding rules, a compressed stream can be obtained: 000010001110010111101101; where the residual coefficient syntax element Coeff_token (the number of non-zero coefficients TotalCoeffs and the number of trailing coefficients Trail ingOnes) is encoded as 0000100.

The encoding of the residual coefficient syntax element Coeff_ token (the number of non-zero coefficients TotalCoeffs and the number of trailing coefficients TrailingOnes) is taken as an example to further illustrate a video entropy coding method proposed by the embodiment of the present invention. The encoding process of the coefficient syntax element Coeff_token.

5201: Scan a 4×4 luma prediction residual block to obtain a residual coefficient syntax element to be encoded; 4×4 luma prediction residual block Zigzag order is mapped to a 16-element data string: 0, 3,

0, 1, -1, -1, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0;

The syntax elements for obtaining residual coefficients to be encoded include: Syntax element Coeff_token (Number of non-zero coefficients (TotalCoeffs) = 5 and number of trailing coefficients (TrailingOnes) = 3);

5202, converting the to-be-coded residual coefficient syntax element into an encoding index Codelndex;

And according to the probability distribution of the to-be-coded residual coefficient syntax element Coeff_token, arranged from largest to smallest, using the coding index Codelndex as the syntax element of the residual coefficient to be encoded (non-zero system) The mapping relationship between the number of numbers TotalCoeffs and the number of trailing coefficients (Trai l ingOnes ); as shown in Table 3, the Coeff_ token and the encoding index of the residual coefficient syntax element to be encoded

Codelndex mapping table;

It can be seen from Table 3 that for the to-be-coded residual coefficient syntax element Coeff_token, the number of non-zero coefficients (TotalCoeffs) = 5 and the number of trailing coefficients (Trai lingOnes) = 3, the coding index

table 3

Trai l ingOnes TotalCoeff Codelndex

0 0 0

1 1 1

2 2 2

3 3 3

0 1 4

1 2 5

3 4 6

2 3 7

3 5 8

0 2 9

1 3 10

2 4 11

3 6 12

0 3 13

1 4 14

2 5 15

3 7 16

0 4 17

1 5 18

2 6 19

3 8 20

0 5 21

1 6 22

2 7 23

3 9 24

0 6 25

1 7 26

2 8 27

3 10 28

0 7 29

1 8 30 2 9 31

0 8 32

0 9 33

1 9 34

2 10 35

3 11 36

0 10 37

1 10 38

2 11 39

3 12 40

0 11 41

1 11 42

2 12 43

3 13 44

0 12 45

1 12 46

2 13 47

3 14 48

0 13 49

1 14 50

2 14 51

3 15 52

0 14 53

1 15 54

2 15 55

3 16 56

0 15 57

1 16 58

2 16 59

0 16 60

1 13 61

S203 converting the code index Codelndex into a codeword value CodeNum;

In this embodiment, the mapping relationship between the coding index and the codeword value is established according to the probability statistical distribution of the syntax element to be encoded and the regular coding; in this example, the Golomb-Rice coding rule is adopted. In fact, this part of the embodiment is a lookup table. Operation, according to the index Codelndex lookup table 4, the codeword value and the regularly encoded parameters (orders, etc.) are obtained.

As shown in Table 4, the mapping relationship between the encoding index Codelndex and the codeword value CodeNum, from the table 4, it can be seen that the embodiment of the present invention adopts a Golomb-Rice code of four order of k=0, 1, 2, 3; as can be seen from Table 4, in the embodiment of the present invention, when the code index is CodeIndex=8, corresponding Code value CodeNum=22, order k=2;

Table 4

84 3 34

85 3 35

86 3 36

87 3 37

88 3 38

89 3 39

90 3 40

91 3 41

92 3 42

93 3 43

94 3 44

95 3 45

96 3 46

97 3 47

98 3 48

99 3 49

100 3 50

101 3 51

102 3 52

103 3 53

104 3 54

105 3 55

106 3 56

107 3 57

108 3 58

109 3 59

110 3 60

111 3 61

S204, the codeword value CodeNum is coded according to the codeword encoding rule, and the final code is obtained according to the codeword value and the codeword encoding rule, and the codeword is generated by the codeword value calculation. The embodiment of the present invention adopts the variable length coding mode. Golomb-Rice of order, CodeNum=22, according to the Golomb-Rice coding rule, the output is: 0000 0110;

As shown in Table 5, the mapping relationship between the codeword value CodeNum and the codeword CodeWord (Golomb-Rice code) is shown in the embodiment of the present invention.

table 5

2 10 0000 0000 0011 01 0000 0000 001101 85 3

3 11 0000 0000 0011 00 0000 0000 001110 86 3

0 10 0000 0000 0010 11 0000 0000 001111 87 3

1 10 0000 0000 0010 10 0000 0000 0001000 88 3

2 11 0000 0000 0010 01 0000 0000 0001001 89 3

3 12 0000 0000 0010 00 0000 0000 0001010 90 3

0 11 0000 0000 0001 111 0000 0000 0001011 91 3

1 11 0000 0000 0001 110 0000 0000 001100 92 3

2 12 0000 0000 0001 101 0000 0000 0001101 93 3

3 13 0000 0000 0001 100 0000 0000 0001110 94 3

0 12 0000 0000 0001 on 0000 0000 0001111 95 3

1 12 0000 0000 0001 010 0000 0000 00001000 96 3

2 13 0000 0000 0001 001 0000 0000 00001001 97 3

3 14 0000 0000 0001 000 0000 0000 00001010 98 3

0 13 0000 0000 0000 1111 0000 0000 00001011 99 3

1 14 0000 0000 0000 1110 0000 0000 00001100 100 3

2 14 0000 0000 0000 1101 0000 0000 00001101 101 3

3 15 0000 0000 0000 1100 0000 0000 00001110 102 3

0 14 0000 0000 0000 1011 0000 0000 00001111 103 3

1 15 0000 0000 0000 1010 0000 0000 0000 01000 104 3

2 15 0000 0000 0000 1001 0000 0000 000001001 105 3

3 16 0000 0000 0000 1000 0000 0000 000001010 106 3

0 15 0000 0000 0000 0111 0000 0000 000001011 107 3

1 16 0000 0000 0000 0110 0000 0000 000001100 108 3

2 16 0000 0000 0000 0101 0000 0000 000001101 109 3

0 16 0000 0000 0000 0100 0000 0000 000001110 110 3

1 13 0000 0000 0000 001 0000 0000 000001111 111 3 In the embodiment of the present invention, by introducing a coding index and a regular variable length/fixed length coding method, a code table of each syntax element is newly designed, and a fast code table retrieval is realized by the coding index. The rule variable length/fixed length coding does not need to save the actual binary codeword, which eliminates the need for the traditional variable length code table to occupy a large memory space and needs to be traversed bit by bit. Under the premise of ensuring coding efficiency, the mobile terminal device is greatly improved. The implementation complexity of the upper entropy decoding operation is suitable for portable terminals with limited hardware resources. Embodiment 3 FIG. 3 is a schematic structural diagram of a video entropy coding apparatus according to an embodiment of the present invention. The apparatus includes an encoding syntax element acquisition module, a first conversion module, a second conversion module, and a first codeword acquisition module.

a coding syntax element obtaining module, configured to acquire a syntax element to be encoded in a block to be encoded; the syntax element to be encoded includes a residual coefficient syntax element or a non-residual coefficient syntax element; the non-residual coefficient syntax element includes a macroblock Type, reference frame index, quantization parameter, motion vector residual, coding model or intra prediction mode;

a first conversion module, configured to convert the syntax element to be encoded into a coding index; and include a probability distribution distribution mapping module;

The probabilistic statistical distribution mapping module is configured to obtain probability statistical distribution information of a syntax element to be encoded in a block to be encoded, perform probability ranking on the to-be-encoded syntax element according to the probability statistical distribution information, and establish the to-be-encoded syntax. The mapping relationship between the element and the encoding index;

Further, the first conversion module further includes a non-probability statistical distribution mapping module, and the non-probabilistic statistical distribution mapping module is configured to convert the non-residual coefficient syntax elements in the syntax element to be encoded according to the non-probabilistic statistical distribution mapping method. An encoding index; the non-probabilistic statistical distribution mapping method includes an unsigned direct mapping, a signed conversion to a symbol mapping or a mapping table manner;

a second conversion module, configured to convert the coding index into a codeword value; the mapping manner of converting the code index Codelndex into a codeword value CodeNum may be a probability statistical distribution map or a non-probabilistic statistical distribution map;

The method for converting the coding index Codelndex into the codeword value CodeNum is to: establish a mapping relationship between the coding index and the codeword value according to the probability statistical distribution information of the syntax element to be encoded; wherein the mapping relationship follows The principle of the shortest average code length, the codeword value and the codeword of the mapped by the probability of the syntax element to be encoded: the short codeword is assigned to the large probability codeword value, and the long codeword is assigned by the small probability codeword value;

The non-probabilistic statistical distribution mapping method of converting the encoding index Codelndex into a codeword value CodeNum is:

Coding the code index Codelndex and the codeword value CodeNum linearly, that is, CodeIndex=axCodeNum+p (where a and β are integers); or establishing a mapping relationship between the coded value and the codeword value according to the mapping table; a first codeword obtaining module, configured to encode the codeword value according to a codeword encoding rule, to obtain a final codeword;

The codeword encoding rule may be a regular variable length coding or a regular fixed length coding;

The regular variable length coding may be a Columbus-Rice or an index Columbus coded Exp-Golomb coding rule of different order; the order is a Columbus coded parameter, and the coding formulas of different orders are different;

In the embodiment of the present invention, by introducing a coding index and a regular variable length/fixed length coding method, each syntax element code table is newly designed, and a fast code table retrieval is realized by the coding index, and the rule variable length/fixed length coding does not need to save the actual binary. The codeword eliminates the need for a large memory space and requires a bit-by-bit traversal operation. Under the premise of ensuring coding efficiency, the implementation complexity of the entropy decoding operation on the mobile terminal device is greatly improved, and is applicable to Portable end of limited hardware resources. Embodiment 4

A recording medium, including the program of the video entropy coding method described in Embodiment 1, the video entropy coding method is described in detail in Embodiment 1, and details are not described herein again. Embodiment 5

FIG. 4 is a flowchart of a video entropy decoding method according to an embodiment of the present invention, where the method includes the following steps:

S501: reading the entropy-encoded binary bit stream, performing calculation according to the codeword encoding rule, and obtaining a complete binary codeword Codeword;

The codeword has a regular codeword structure, the codeword value CodeNum and the codeword CodeWord There is a certain mathematical relationship directly, and the codeword can be generated by calculating the codeword value, and the codeword value can also be calculated by the codeword;

The regular variable length coding may be a Columbus-Rice or exponential Columbus coded Exp-Golomb coding rule of different order; the order is a Columbus coded parameter, and the coding formulas of different orders are different;

The "reading the entropy-encoded binary bit stream, calculating according to the codeword encoding rule, and obtaining the complete binary codeword Codeword"

Specifically:

A) reading the entropy encoded binary bit stream;

B) obtaining the number of "0"s before the first "1" in the binary bit stream, and using the number as the prefix information in the codeword structure;

Obtaining a complete binary codeword according to the prefix information and the codeword encoding rule; for example: a prefix of a codeword in a binary bitstream is in the form of 1, 01, 001, 0001, 00001, 000001, 0000001, 00000001, 000000001, 0000000001, 00000000001, 000000000001, 0000000000001, 00000000000001 When decoding, the first "1" detector method is generally used. Specifically, the binary bit stream is input, and a sequence of 0 plus one is detected first. Count the number of 0s. Such detectors are often named the first "1" detector. After the first "1" is detected, the number of the first 0 is assigned to the variable M_prefix to represent the prefix information. It can be concluded that the value of the variable M_prefix is 0, 1, 2, 3, 4 respectively. , 5, 6, 7, 8, 9, 10, 11, 12, 13.

S502: The binary codeword Codeword is calculated according to the codeword encoding rule to obtain a codeword value CodeNum;

S503: Convert the codeword value CodeNum into an encoding index Codelndex;

The mapping method for converting the codeword value CodeNum into the encoding index Codelndex may be a probability statistical distribution mapping method or a non-probabilistic statistical distribution mapping method;

Converting the codeword value CodeNum into a probabilistic statistical distribution map of the encoding index Codelndex The method is: establishing a mapping relationship between the coding index and the codeword value according to the probability statistical distribution information of the syntax element to be encoded; wherein the mapping relationship follows the principle of the shortest average code length, and the code mapped by the probability of the syntax element to be encoded Word value and codeword: The large probability codeword value is assigned a short codeword, and the small probability codeword value is assigned a long codeword;

The non-probabilistic statistical distribution mapping method for converting the codeword value CodeNum into the encoding index Codelndex is:

S504: Obtain a corresponding syntax element according to the code index Codeindex;

The syntax element to be encoded includes a residual coefficient syntax element or a non-residual coefficient syntax element. The embodiment of the present invention provides a low complexity entropy decoding method, which is constructed according to H. 264 CAVLC listing context adaptation. The multiple VLC tables, the embodiment of the present invention only needs to control the mapping relationship between the code index Codelndex and the codeword value CodeNum, so that the effect of multiple VLC tables can be achieved, and the regular variable length/fixed length coding does not need to save the actual binary code words. CodewWord, saving codeword data and encoding index memory requirements is lower than saving binary codeword Codeword, saving resources. Embodiment 6

The following is a process step of entropy decoding to obtain a syntax element Coeff_token by entropy decoding of the binary bit stream after entropy coding in the embodiment of the present invention:

S601, reading an entropy-encoded binary bit stream, and acquiring a complete binary codeword according to Golomb's codeword encoding rule;

In the embodiment of the present invention, the first "1" detector is used to read the codeword of the compressed bit stream, and the legal codeword generally has a string of 0 plus a prefix feature of 1, so that the first "1" can be detected. The number of the preceding 0 is assigned to the variable M_prefix, and then the entire binary codeword is obtained by the encoding rule of Golomb; the variable M_prefix is used to represent the prefix information in the codeword structure;

If the entropy-encoded binary bit stream is 0000 Ol lOxxxxx, the number of "0"s before the first "1" in the binary bit stream is obtained by detecting the binary bit stream, the "0" The number is 5, and 5 is assigned to the variable M_prefix, M_prefix=5. According to the encoding rule of Golomb_Rice, the order k=2 (4<M—prefix<=9) is determined to determine the complete codeword. The form is 0000 Olxx, the length is 8, so continue to shift to read the 8-bit binary string: 0000 0110.

S602, according to the coding rule of Golomb-Rice and the order k=2, calculate the codeword value CodeNum=22;

5603. According to the mapping relationship between the code index Codelndex and the codeword value CodeNum (as shown in Table 4 in Embodiment 2), when the codeword value CodeNum=22, the code index CodeIndex=8 is obtained.

5604, according to the mapping table of the residual coefficient syntax element Coeff_token and the encoding index Codelndex (as shown in Table 3 in Embodiment 2), when the encoding index Codeindex=8, the corresponding syntax element Coeff_token is obtained: The number of zero coefficients (TotalCoeff s ) = 5 and the number of trailing coefficients (Trai l ingOnes) = 3.

Compared with the decoding process of H.264 CAVLC, since the H.264 code table code word has no regularity, after reading bit by bit from the compressed bit stream, it is necessary to traverse the VLC table. If this is an n-bit code word, this requires 62η. The data of the variable length code table VLC is read in order to confirm the complete code word, for example, the code word is 7 bits, which requires 7 X 62=432 table lookup operations, which will inevitably consume a lot of computing resources, repeated Read and write operations also increase power consumption. The embodiment of the present invention provides a low-complexity entropy decoding method, and the embodiment of the present invention only needs to control the encoding index Codelndex and the codeword value compared to the H.264 CAVLC listing the multiple VLC tables constructed by the context adaptation. The mapping between CodeNum can achieve the effect of multiple VLC tables, and there is no need to save the actual binary codeword CodewWord with regular variable length/fixed length encoding. The memory requirement for saving codeword data and encoding index is lower than saving binary codeword CodeWord. More resource saving. Example 7

The following is an embodiment of the present invention for entropy-encoded binary bit stream, using Golomb encoding rules for entropy decoding to obtain the number of zeros of the non-zero coefficients of the syntax element Total-zeros:

Table 6 (including Table 6a and Table 6b) shows the mapping relationship between the syntax elements Total_zeros, the code index CodeIndex, and the codeword value CodeNum in the embodiment of the present invention.

To facilitate decoding, the embodiment of the present invention sets the encoding index Codelndex equal to the syntax element Total_zeros, and uses the Golomb-Rice code as the encoding rule. 5701, reading an entropy-encoded binary bit stream, and acquiring a complete binary codeword according to a coding rule of Golomb-Rice;

In the embodiment of the present invention, the first "1" detector is used to read the codeword of the compressed bit stream, and the legal codeword generally has a string of 0 plus a prefix feature of 1, so that the first "1" can be detected. The number of the preceding 0 is assigned to the variable M_prefix, and the entire binary codeword is obtained by the encoding rule of Golomb; the variable M_prefix is used to represent the prefix information in the codeword structure;

If the entropy-encoded binary bit stream is OlOlOxxxxx, the number of the "0"s before the first "1" in the binary bitstream is obtained by detecting the binary bitstream. For 1, assign 1 to the variable M_prefix, M_prefix=l, according to the coding rule of Golomb-Rice and the order k=2 (TotalCoeffs is equal to 5), to determine that the complete codeword form is Olxx, read Take the 4-digit codeword Codeword: 0101;

5702, according to the encoding rule of Golomb-Rice and the order k=2, according to the calculated codeword value CodeNum=5;

5703, according to the mapping relationship of Table 6 (a), when the codeword value CodeNum=5, the coded bow I CodeIndex=2;

5704, the syntax element Total_zeros is equal to 2 according to the encoding index Codeindex=2; in the embodiment of the present invention, the encoding index Codelndex is equal to the syntax element Total_zeros, so the value of the syntax element Total_zeros can be directly obtained according to the value of the encoding index. ; Table 6a mapping relationship between the syntax element Total_zeros, the encoding index CodeIndex, and the codeword value CodeNum

■4^

*3⁄43⁄47t ! 6botaeos Codede Codeu Tl zrInxNm

In the embodiment of the present invention, due to the regular structure of the codeword, the complete codeword can be quickly determined without the need for bit-by-bit reading of the traditional variable length code table and cumbersome operation of traversing the VLC table, thereby reducing the calculation of entropy decoding. Complexity, suitable for resource-constrained applications. This code table can be improved by combining the first "1" detection method to optimize the codeword. For example, 0001 10 is the longest and only codeword of this length, so it can be reached by judging the "0" after the number "1" is ignored. Further refine the effect of the codeword to obtain a better compression ratio. Embodiments of the present invention provide a low complexity entropy decoding method, compared to

H. 264 CAVLC lists multiple VLC tables constructed by context adaptation. In this embodiment of the present invention, only the mapping relationship between the code index Codelndex and the codeword value CodeNum can be controlled, and the effect of multiple VLC tables can be achieved, and the rules are changed. The long/fixed length encoding does not need to save the actual binary codeword CodewWord. The memory requirement for saving the codeword data and the encoding index is lower than saving the binary codeword Codeword, which saves resources. Example eight

FIG. 5 is a schematic structural diagram of a video entropy decoding apparatus according to an embodiment of the present invention. The apparatus includes: a complete codeword acquisition module, a codeword value acquisition module, a third conversion module, and a second syntax element acquisition module.

a complete codeword obtaining module, configured to read the entropy-encoded binary bit stream, perform calculation according to the codeword encoding rule, and obtain a complete binary codeword;

The complete codeword acquisition module includes a binary bit stream reading module and a codeword prefix information acquisition module. Block and second codeword acquisition module,

a binary bit stream reading module, configured to read the entropy encoded binary bit stream; a codeword prefix information obtaining module, configured to acquire a number of “0”s before the first “1” in the binary bit stream, The number is used as prefix information in the codeword structure;

a second codeword obtaining module, configured to obtain a complete binary codeword according to the prefix information and the codeword encoding rule;

The mapping manner of converting the codeword value CodeNum into the encoding index Codelndex may be a probability statistical distribution mapping manner or a non-probabilistic statistical distribution mapping manner;

The probability distribution distribution mapping manner of converting the codeword value CodeNum into the coding index Codelndex is: establishing a mapping relationship between the coding index and the codeword value according to a probability statistical distribution of the syntax element to be encoded; wherein the mapping relationship follows an average The principle of the shortest code length, the codeword value and the codeword of the mapped by the probability of the syntax element to be encoded: the short codeword is assigned to the large probability codeword value, and the long codeword is assigned by the small probability codeword value;

The non-probabilistic statistical distribution mapping method of converting the codeword value CodeNum into the encoding index Codelndex is:

The second syntax element obtaining module is configured to obtain a corresponding syntax element according to the encoding index. The embodiment of the present invention provides a low complexity entropy decoding method, which is compared with the 上下文. 264 CAVLC listing context adaptive The VLC table, the embodiment of the present invention only needs to control the mapping relationship between the code index Codelndex and the codeword value CodeNum, so that the effect of multiple VLC tables can be achieved, and the regular variable length/fixed length coding does not need to save the actual binary codeword CodewWord. Saving memory for codeword data and encoding index is lower than saving binary codeword Codeword, which saves resources. Example nine

A recording medium, including the program of the video entropy decoding method described in Embodiment 5, the video entropy decoding method is described in detail in Embodiment 5, and details are not described herein again.

It will be understood by those skilled in the art that all or part of the steps of the foregoing embodiments may be implemented by a program instruction related hardware, and the program may be stored in a computer readable storage medium. The storage medium may be a ROM, a RAM, a magnetic disk, an optical disk, or the like.

The above is only the preferred embodiment of the present invention, and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection of the present invention. Within the scope.

Claims

Rights request

A video entropy coding method, the method comprising:

Obtaining a syntax element to be encoded in the block to be encoded;

Converting the syntax element to be encoded into a coding index;

Converting the encoded index into a codeword value;

The video entropy coding method according to claim 1, wherein the to-be-encoded syntax element comprises: a residual coefficient syntax element or a non-residual coefficient syntax element.

The video entropy coding method according to claim 2, wherein the method for obtaining the residual coefficient syntax element in the block to be coded is as follows:

Recalculating the residual coefficients of the block to be coded after the prediction is performed in a scanning order, and arranging the two-dimensional residual coefficients into a one-dimensional sequence;

Obtaining a residual coefficient syntax element of the block to be coded according to a statistical feature of the residual coefficient in the one-dimensional sequence.

The video entropy coding method according to claim 2, wherein the method for converting a syntax element to be encoded into a coding index comprises: a probability statistical distribution mapping method or a non-probabilistic statistical distribution mapping method.

5. The video entropy encoding method according to claim 4, wherein

When the method for converting a syntax element to be encoded into a coding index is a probability statistical distribution mapping method, the method for converting the syntax element to be encoded into an encoding index is specifically:

Obtaining probability statistical distribution information of the syntax element to be encoded in the block to be coded; performing probability ranking on the syntax element to be encoded according to the probability distribution information;

Establishing a mapping relationship between the syntax element to be encoded and the encoding index;

When the method for converting a syntax element to be encoded into a coding index is a non-probabilistic statistical distribution mapping method, the method for converting the syntax element to be encoded into an encoding index is specifically:

Unsigned direct mapping, signed to symbolic mapping or mapping table mode.

The video entropy coding method according to claim 5, wherein the obtaining probability statistical distribution information of the syntax element to be encoded in the block to be encoded is specifically:

Obtaining a predictive quantized residual coefficient of the block to be coded;

And calculating a probability statistical distribution value of the syntax element based on the context model according to the to-be-coded block residual coefficient, as the probability statistical distribution information of the syntax element to be encoded.

The video entropy coding method according to claim 6, wherein the "statistical statistical distribution value of a syntax element based on a context model is calculated according to the residual coefficient of the block to be encoded as the syntax element to be encoded. The probabilistic statistical distribution information" also includes:

The probabilistic statistical distribution information of the probabilistic statistical distribution value is obtained, and the probability statistical distribution value of the merged syntax element is obtained as the probability statistical distribution information of the syntax element to be encoded.

The video entropy coding method according to claim 5, wherein the establishing a mapping relationship between the to-be-encoded syntax element and the coding index is specifically: establishing the non-residual coefficient syntax element and the The mapping of the encoded index is to establish a mapping of a single non-residual coefficient syntax element to the encoding index, or/and to establish a joint mapping of a plurality of mappings of non-residual coefficient syntax elements and the encoding index.

The video entropy coding method according to claim 8, wherein the "joining a joint mapping of a plurality of mappings of non-residual coefficient syntax elements and the coding index" is to integrate at least two non-residual coefficients The joint probability distribution of the syntax elements is used to establish a corresponding non-residual coefficient syntax element mapping table, or the non-residual coefficient syntax elements are grouped and then comprehensively encoded.

The video entropy coding method according to claim 1, wherein the mapping method for converting the coding index into a codeword value is a probability statistical distribution mapping method.

The video entropy coding method according to claim 10, wherein the mapping method for converting the coding index into a codeword value is specifically:

And establishing a mapping relationship between the encoding index and the codeword value according to probability statistical distribution information of a syntax element to be encoded.

The video entropy coding method according to claim 1, wherein the mapping method for converting the coding cable into a codeword value is a non-probabilistic statistical distribution mapping method.

The video entropy coding method according to claim 12, wherein the mapping method for converting the coding index into a codeword value is specifically: And linearly associating the coding index with the codeword value, or establishing a mapping relationship between the coding cable and the codeword value by formulating a mapping table;

The linear correlation formula is: Coded Ι = α χ code word value + β, where α and β are integers.

The video entropy coding method according to claim 1, wherein the codeword coding rule comprises a regular variable length coding or a regular fixed length coding.

The video entropy encoding method according to claim 14, wherein the regular variable encoding comprises Columbus or index Columbus encoding of different orders.

16. A video entropy coding apparatus, the apparatus comprising:

a coding syntax element obtaining module, configured to obtain a syntax element to be encoded in the block to be coded; a first conversion module, configured to convert the syntax element to be encoded into a coding index;

The video entropy coding apparatus according to claim 16, wherein the first conversion module comprises a probability statistical distribution mapping module.

The probabilistic statistical distribution mapping module is configured to obtain probability statistical distribution information of a syntax element to be encoded in a block to be encoded, perform probability ranking on the to-be-encoded syntax element according to the probability statistical distribution information, and establish the to-be-encoded syntax. The mapping relationship between the element and the encoding index.

The video entropy coding apparatus according to claim 16, wherein the first conversion module further comprises a non-probabilistic statistical distribution mapping module.

The non-probabilistic statistical distribution mapping module is configured to convert a non-residual coefficient syntax element in a syntax element to be encoded into a coding index according to a non-probabilistic statistical distribution mapping method; the non-probabilistic statistical distribution mapping method includes an unsigned direct mapping, Signed into a symbol map or map table.

The video entropy coding apparatus according to claim 16, wherein the mapping manner of converting the coding index into a codeword value is a probability statistical distribution map, and converting the coding index into a codeword value The probability distribution distribution mapping manner is specifically: establishing a mapping relationship between the coding index and the codeword value according to the probability statistical distribution information of the syntax element to be encoded.

The video entropy coding apparatus according to claim 16, wherein the mapping manner of converting the coding index into a codeword value is a non-probabilistic statistical distribution map, and converting the coding index into a codeword value The non-probabilistic statistical distribution mapping method is specifically as follows:

And linearly associating the coding index with the codeword value, or establishing a mapping relationship between the coding cable and the codeword value by formulating a mapping table;

The video entropy encoding apparatus according to claim 16, wherein the codeword encoding rule comprises a regular variable length coding or a regular fixed length coding.

A recording medium comprising the program of the video entropy encoding method according to any one of claims 1 to 15.

A video entropy decoding method, the method comprising:

Reading the entropy-encoded binary bit stream, performing calculation according to the codeword encoding rule, and obtaining a complete binary codeword;

Converting the codeword value into an encoding index;

Corresponding syntax elements are obtained according to the encoding.

The video entropy decoding method according to claim 23, wherein the reading the entropy encoded binary bit stream is performed according to a codeword encoding rule to obtain a complete binary codeword.

Reading the entropy encoded binary bit stream;

Obtaining a number of “0”s before the first “1” in the binary bit stream, and using the number as the prefix information in the codeword structure;

Obtaining a complete binary codeword according to the prefix information and the codeword encoding rule.

The video entropy decoding method according to claim 23, wherein the mapping method for converting the codeword value into a coding index comprises a probability statistical distribution mapping method or a non-probabilistic statistical distribution mapping method.

The video entropy decoding method according to claim 25, wherein

When the mapping method for converting the codeword value into a coding index is a probability statistical distribution mapping In the method, the mapping method for converting the codeword value into the coding function is as follows: establishing a mapping relationship between the coding index and the codeword value according to probability statistical distribution information of a syntax element to be encoded;

When the mapping method for converting the codeword value into a coding index is a non-probabilistic statistical distribution mapping method, the mapping method for converting the codeword value into an encoding index is specifically:

The linear correlation formula is: Encoding index = α χ code word value + β, where α and β are integers.

27. A video entropy decoding apparatus, the apparatus comprising:

28. The video entropy decoding apparatus according to claim 27, wherein:

The complete codeword acquisition module includes a binary bitstream reading module, a codeword prefix information acquisition module, and a second codeword acquisition module.

The binary bit stream reading module is configured to read the entropy-encoded binary bit stream; the codeword prefix information obtaining module is configured to acquire the first “1” before the “0” in the binary bit stream The number of the number, the number is used as the prefix information in the codeword structure;

The second codeword obtaining module is configured to obtain a complete binary codeword according to the prefix information and the codeword encoding rule.

The video entropy decoding apparatus according to claim 27, wherein the mapping method for converting the codeword value into a coding index comprises a probability statistical distribution mapping method or a non-probabilistic statistical distribution mapping method.

The video entropy decoding apparatus according to claim 29, wherein when the mapping method for converting the codeword value into a coding index is a probability statistical distribution mapping method, the code is The mapping method for converting a word value into a coding index is specifically as follows:

Establishing a mapping relationship between the encoding index and the codeword value according to probability statistical distribution information of a syntax element to be encoded;

A recording medium comprising the program of the video entropy decoding method according to any one of claims 23 to 26.