CN100423582C - A method and device for binary coding data to be coded - Google Patents

Abstract

The present invention discloses a method and a device for binary coding to data to be coded, which carries out the binary coding to the data to be coded. Firstly, ranges of the data to be coded are divided into a plurality of subsidiary ranges, each subsidiary range is provided with a subsidiary range mark bit string and a subsidiary range base value which are corresponding to the subsidiary range; then, the subsidiary ranges of the data to be coded can be determined according to the magnitude of the numerical value of the data to be coded, and the corresponding subsidiary range mark bit string and the subsidiary range base value can be generated simultaneously; subsequently, a subsidiary range relative value can be obtained by using the data to be coded to subtract the subsidiary range base value, and a corresponding binary coding mechanism is adopted to carry out the binary coding to the obtained subsidiary range relative value according to the generated subsidiary range mark bit string; thereby, a subsidiary range code can be obtained, and the subsidiary range code is spliced to the rear part of the subsidiary range mark bit string. The high coding efficiency can be achieved by combining the binary coding method and a binary arithmetic coding method. The present invention is used for the field of digital signal processing, particularly the field of video coding and decoding or image coding and decoding.

Description

A method and device for binary coding data to be coded

technical field

The invention relates to an electric digital data processing technology, in particular to a method and device for binarizing data to be encoded.

Background technique

Entropy coding is widely used in video coding and image coding, and its function is to remove the statistical redundancy of video data or image data, so as to achieve the effect of data compression.

Entropy coding can be roughly divided into two categories: Huffman coding and arithmetic coding. Since the Huffman code is based on the static statistics of the source data distribution, it is difficult to adaptively encode the information source according to the conditional probability, so the conditional redundant information between the source data cannot be well represented. removed, thus limiting its compression performance. Arithmetic coding can conveniently carry out self-adaptive dynamic statistics on source distribution, can well apply conditional information between source data, and remove conditional correlation between source data, so it has a better performance than Huffman code. Compression performance, but its algorithm complexity is relatively high.

Context-based Adaptive Binary Arithmetic Coding (CABAC) adopts binary arithmetic coding, which makes the probability estimation of source symbols simple. According to statistics, it achieves better compression performance than Huffman code, so this algorithm has been adopted by the international video coding standard H.264.

Figure 1 shows a coding block diagram of context-based adaptive binary arithmetic coding (CABAC). It can be seen from the figure that CABAC coding has three technical links: binarization of syntax elements, context-based probability modeling and binary arithmetic coding engine. Among them, the binarization of grammatical elements is a key link in CABAC. A reasonable and effective binarization algorithm can facilitate the establishment of probability models, and can make full use of the probability distribution characteristics of source symbols, so that CABAC can efficiently analyze source symbols. The data is compressed. Therefore, reasonable and efficient binarization of source symbols plays a vital role in CABAC.

The binarization algorithm used in the H.264 video standard is a combination of Truncated Unary and Exp-Golomb binary mechanisms. It first sets a threshold value. When the value to be encoded is lower than this threshold value, Use the Truncated Unary binary mechanism to binarize the encoded data; when the value to be encoded is higher than or equal to the threshold value, first, generate a longer prefix code, and then calculate the difference between the encoded data and the threshold value The Exp-Golomb binary mechanism is used for binarization, and the binarized result is connected to the prefix code to generate the final binary representation of the data to be encoded. Its disadvantage is that for data with large values, a large part of its binarization results are prefix codes, which is not conducive to compression.

Introduction to Truncated Unary codes and Exp-Golomb codes

Truncated Unary code is a relatively simple code word, which is mainly used to process data with an upper limit in the value range. Suppose the data to be encoded is x, the upper limit of the value range is s, where x and s are both non-negative integers, if x<s, then the binarization result of x is x prefix bits b0, and then ends with a b1 bit, where b0, b1∈{0, 1}, and b0≠b1; if x==s, the binarization result of x is x prefix bits b0. Example: For a Truncated Unary code with an upper limit value of 8, the encoding result of 5 is b0 b0 b0 b0 b0 b0 b1, and the encoding result of 8 is b0 b0 b0 b0 b0 b0 b0 b0 b0 b0 b0.

The Exp-Golomb code is more complicated than the Truncated Unary code, and the Exp-Golomb code has a parameter k to define its order. The structure of the 0-order, 1-order, 2-order and 3-order exponential Golomb codes is given in the following table, and x _i ∈ {0, 1} in the table.

Contents of the invention

The object of the present invention is to provide a method and device for binary coding data to be coded. After the data to be encoded is processed by the binarization algorithm of the present invention, the probability distribution characteristics of the source data can be well reflected, and after the data to be encoded is processed by the binarization of the present invention, the probability estimation of the source data is convenient and accurate In this way, the present invention is combined with binary arithmetic coding, and can achieve high-efficiency compression performance for video signals, such as transform coefficients, motion vectors and other information to be coded. Furthermore, the present invention is also applicable to huffman codes.

In order to achieve the above object, the technical scheme adopted in the present invention is as follows:

One, a kind of method that is used to carry out binarization encoding to the data to be encoded, the steps of this method are as follows:

a) Divide the value range of the data to be encoded into m subfields, each subfield has a corresponding subfield flag bit string and a subfield base value;

b) According to the numerical value of the data to be encoded, determine the subfield to which it belongs, and simultaneously generate the corresponding subfield flag bit string and subfield base value;

c) Subtract the base value of the subfield from the data to be encoded to obtain the relative value of the subfield;

d) According to the sub-field flag bit string generated in step b), use the corresponding binarization coding mechanism to perform binarization coding on the relative value of the sub-field obtained in step c), thereby obtaining the sub-field code, different The subfield flag bit string corresponds to different binarization coding mechanisms;

e) After the subfield code is spliced into the subfield flag bit string, the binary representation of the data to be encoded is obtained.

For the m subfields, m is equal to 2, the length of the subfield flag bit string is 1 bit, the subfield flag bit string of the first subfield is b0, the subfield base value is bv0, and the subfield flag of the second subfield is b0. The bit string is b1, the base value of the subfield is bv1, where b0, b1∈{0, 1}, and b0≠b1, bv0<bv1.

Determine the subfield to which the data to be encoded belongs as follows: if the data to be encoded is greater than or equal to the subfield base value bv0 and the data to be encoded is less than the subfield base value bv1, then the data to be encoded belongs to the first subfield; if the data to be encoded is greater than or equal to the subfield If the field base value is bv1, then the data to be encoded belongs to the second subfield.

For the data to be encoded whose subfield flag bit string is b0, use the Truncated Unary binary encoding mechanism with a truncated value of bv1-bv0-1 to binarize the relative value of its subfield; for the data to be encoded whose subfield flag bit string is b1 For the encoded data, the k-order Exp-Golomb binarization coding mechanism is used to binarize the relative value of its subfields, where k is a non-negative integer.

Two, a kind of device that is used to carry out binarization encoding to the data to be encoded, it comprises:

1) The value range of the data to be encoded can be divided into several subfields, the identification bit string of each subfield and the base value of the subfield can be determined, and the corresponding subfield identification bit string and the relative value of the subfield can be output according to the numerical value of the data to be encoded Sub-domain dividing means;

2) According to the input subfield flag bit string, use the corresponding binarization mechanism to binarize the input subfield relative value to obtain the subfield code, and finally splice the subfield code into the subfield flag bit string to obtain the code to be coded A subfield code encoding device for binary representation of data;

The input end of the subfield code encoding device is connected with the data to be encoded, and the subfield mark bit string output terminal and the subfield relative value output end of the subfield code encoding device are respectively connected to the input end of the subfield code encoding device, and the subfield code encoding device's The output is the output of the binarized representation of the data to be encoded.

The beneficial effects that the present invention has are:

The binarization of the data to be encoded according to the above method can not only better reflect the distribution characteristics of the data to be encoded, but also adopt the corresponding binarization mechanism according to the probability distribution characteristics of the data to be encoded in different sub-domains. Therefore, the probability estimation of the data to be coded is more accurate, and combined with the binary arithmetic coding, the coding and decoding efficiency can be greatly improved. The invention is applied in the field of digital signal processing, especially in the field of video codec or image codec.

Description of drawings

Figure 1 is a coding block diagram of context-based adaptive binary arithmetic coding;

Fig. 2 is a block diagram of an encoding device of the present invention.

Detailed ways

The present invention will be further described below in conjunction with drawings and embodiments.

Embodiment 1: Binarization of the Level absolute value of the transformation coefficient

Accompanying drawing 2 shows the binarized coding device of the present invention, including the sub-field division device and the code synthesis device.

The currently to-be-encoded data (denoted as data_value) is input to the sub-field division device. The sub-field division device divides the value domain of the data to be encoded into m=2 sub-fields, the corresponding sub-field base values are respectively bv0=1 and bv1=15, and the sub-field flag bit strings are respectively b0='0' and b1=' 1', that is, if the data to be encoded is greater than 0 and less than 15 at the same time, the subfield division device outputs the subfield flag bit string '0' and the relative value data_value-1 of the subfield; if the data to be encoded is greater than or equal to 15, the subfield division device outputs the subfield Field flag bit string '1' and subfield relative value data_value-15.

The subfield flag bit and the subfield relative value are input to the subfield code coding device, and the subfield code coding device uses a corresponding binarization mechanism to perform binarization coding on the subfield relative value according to the subfield flag bit. That is, if the subfield flag bit string is '0', the Truncated Unary coding mechanism is used to binarize the relative value of the subfield; if the subfield flag bit string is '1', the k=0-order Exp-Golomb coding mechanism is used to encode the subfield Domain relative values are encoded in binarization.

Finally, the subfield flag bit string and the bit string generated after the binarization of the subfield relative value are connected in parallel. When connecting in parallel, the subfield flag bit string is first, and the bit string generated after the subfield relative value is binarized is behind. , and finally obtain the binary representation of the data to be encoded.

Embodiment 2: Binarization of the absolute value of the motion vector residual component

Accompanying drawing 2 shows the binarized coding device of the present invention, including the sub-field division device and the code synthesis device.

The currently to-be-encoded data (denoted as data_value) is input to the sub-field division device. The subfield division device divides the data value domain to be encoded into m=2 subfields, the corresponding subfield base values are respectively bv0=0 and bv1=9, and the subfield flag bit strings are respectively b0='0' and b1=' 1', that is, if the data to be encoded is greater than or equal to 0 and less than 9 at the same time, the subfield division device outputs the subfield flag bit string '0' and the relative value data_value of the subfield; if the data to be encoded is greater than or equal to 9, the subfield division device outputs the subfield Flag bit string '1' and subfield relative value data_value-9.

The subfield flag bit string and the subfield relative value are input to the subfield code coding device, and the subfield code coding device uses a corresponding binarization mechanism to perform binarization coding on the subfield relative value according to the subfield flag bit string. That is, if the subfield flag bit string is '0', the Truncated Unary encoding mechanism is used to binarize the subfield relative value; if the subfield flag bit string is '1', the k=3rd-order Exp-Golomb coding mechanism is used to encode the Domain relative values are encoded in binarization.

Finally, the subfield flag bit string and the bit string generated after the binarization of the relative value of the subfield are connected in parallel. When concatenated, the subfield flag bit is first, and the bit string generated after the binarization of the relative value of the subfield is behind.

The above-mentioned embodiments are used to illustrate the present invention, rather than to limit the present invention. Within the spirit of the present invention and the protection scope of the claims, any modification and change made to the present invention will fall into the protection scope of the present invention.

Claims (5)

1. method that is used for data to be encoded are carried out binary scale coding is characterized in that the step of this method is as follows:

A) codomain with data to be encoded is divided into m subdomain, and each subdomain all has corresponding with it subdomain flag bit string and subdomain base value;

B) according to the numerical values recited of data to be encoded, determine its affiliated subdomain, generate correspondingly subdomain flag bit string and subdomain base value simultaneously;

C) deduct the subdomain base value with data to be encoded and obtain the subdomain relative value;

D) according at b) the subdomain flag bit string that generated in the step adopts corresponding binary scale coding mechanism to come c) resulting subdomain relative value is carried out binary scale coding in the step, thereby obtain the subdomain sign indicating number, the binary scale coding mechanism that different subdomain flag bit strings is corresponding different;

E) the subdomain sign indicating number being spliced to the binarization that obtains data to be encoded after the subdomain flag bit string represents.

2. a kind of method that is used for data to be encoded are carried out binary scale coding according to claim 1 is characterized in that: a described m subdomain, and its m equals 2, subdomain flag bit string length is 1 bit, the subdomain flag bit string of first subdomain is b0, and the subdomain base value is bv0, and the subdomain flag bit string of second subdomain is b1, the subdomain base value is bv1, wherein, b0, b1 ∈ 0,1}, and b0 ≠ b1, bv0＜bv1.

3. a kind of method that is used for data to be encoded are carried out binary scale coding according to claim 2, it is characterized in that: determine the affiliated subdomain of data to be encoded as follows, if less than subdomain base value bv1, then data to be encoded belong to first subdomain to data to be encoded more than or equal to subdomain base value bv0 and data to be encoded; If data to be encoded are more than or equal to subdomain base value bv1, then data to be encoded belong to second subdomain.

4. a kind of method that is used for data to be encoded are carried out binary scale coding according to claim 3, it is characterized in that: for subdomain flag bit string is the data to be encoded of b0, and the employing cutoff value is that the Truncated Unary binary scale coding mechanism of bv1-bv0-1 is carried out binary scale coding to its subdomain relative value; For subdomain flag bit string is the data to be encoded of b1, adopts k rank Exp-Golomb binary scale coding mechanism that its subdomain relative value is carried out binary scale coding, and wherein k is a nonnegative integer.

5. device that is used for data to be encoded are carried out binary scale coding is characterized in that it comprises:

1) codomain of data to be encoded can be divided into several subdomains, determine each subdomain flag bit string and subdomain base value, according to the corresponding with it subdomain flag bit string of the numerical values recited output of data to be encoded and the subdomain classification apparatus of subdomain relative value;

2) can be according to the subdomain flag bit string of input, adopt corresponding binarization mechanism that the subdomain relative value of input is carried out binary scale coding to obtain the subdomain sign indicating number, the subdomain sign indicating number is spliced to the subdomain code encoding device that the binarization that obtains data to be encoded after the subdomain flag bit string is represented at last;

The input of subdomain classification apparatus links to each other with data to be encoded, the subdomain flag bit string output of subdomain classification apparatus and subdomain relative value output are connected respectively to the input of subdomain code encoding device, and the output of subdomain code encoding device is the output that the binarization of data to be encoded is represented.

Images