CN101646086A - Method and device of preamble reference modelling of 4*4 block conversion coefficient indication - Google Patents
Method and device of preamble reference modelling of 4*4 block conversion coefficient indication Download PDFInfo
- Publication number
- CN101646086A CN101646086A CN 200910168826 CN200910168826A CN101646086A CN 101646086 A CN101646086 A CN 101646086A CN 200910168826 CN200910168826 CN 200910168826 CN 200910168826 A CN200910168826 A CN 200910168826A CN 101646086 A CN101646086 A CN 101646086A
- Authority
- CN
- China
- Prior art keywords
- index value
- block
- conversion coefficient
- preamble
- reference model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Abstract
The invention provides a method of preamble reference modelling of 4*4 block conversion coefficient indication. The method of the invention comprises the following steps: a. obtaining the binary digital state value of 4*4 block end indication by duality; b. combining coded preamble data into the index value of a preset reference model, and selecting the reference model corresponding to the block end indication according to the index value, wherein, the reference model contains a suitable coding parameter when the block end indication codes. The invention also provides a device of preamble reference modelling of 4*4 block conversion coefficient indication. The method/device can effectively improve the coding efficiency of video images and save needed storage space.
Description
Technical field
The present invention relates to encoding video pictures, particularly relate to the method and the device of a kind of 4 * 4 block conversion coefficient indication preamble reference modellings (Context Modeling).
Background technology
The compression of video image now and coding and decoding technology, image be interior or interframe survey prognosis through frame, and the residual error of gained is compressed through conversion again.For example, ITU-T is (Part 10) Advanced Video Coding (hereinafter to be referred as H.264/AVC) and Microsoft WindowsVideo 9 (WMV 9) H.264/MPEG-4, the both adopts 4 * 4 conversion to come the compressed video signal, and AVS (AudioVideo Coding Standard, audio frequency and video coding and decoding technical standard) also considers to adopt 4 * 4 conversion, the importance of visible 4 * 4 conversion.In ABT (Arbitrary Block-size Transform, self adaptation size conversion) system, can adopt the conversion of different sizes, for example 4 * 4,8 * 8 or 16 * 16 conversion.Coefficient after the conversion is again through quantizing the coefficient meeting vanishing after partly quantizing, even all vanishing.More efficient for the order coding, generally use some grammers and represent whether non-zero transform coefficient exists.For example, H.264/AVC and the AVS nonzero coefficient that all adopts Coded Bit Pattern (CBP) to indicate some 8x8 fritters whether exist.But CBP is not enough to represent the situation of 4 * 4 conversion, so will look for method in addition, wherein more direct method is used block end sign (End of Block, EOB) expression of 4 * 4 fritters exactly.When adopting 4 * 4 conversion, before transmitting non-zero transform coefficient, can transmit an EOB earlier, be used for representing whether to have after a while the coefficient of non-zero.Suppose that 4 current * 4EOB represents with X, work as X=0, the coefficient of representing this 4 * 4 fritter is zero entirely; Work as X=1, represent that then this 4 * 4 fritter comprises the coefficient of non-zero.On the other hand, every kind of grammer all must produce code stream behind entropy coding.Entropy coder can adopt CAVLC (Context-AdaptiveVariable-Length Coding, adaptability changes the length coding method) or CABAC (Context-Adaptive Binary Arithmetic Coding, preamble reference adaptive binary arithmetic coding), wherein the efficient of CABAC is higher.Conditional probability in the EOB coding parameter is very important for the CABAC coding, in known AVS CABAC, uses global statistics (conversion coefficient of encoding in same frame) to predict the conditional probability of EOB.
In addition, H.264/AVC standard has proposed encoding block sign (coded_block_flag), this sign is used to indicate the existence of 4 * 4 conversion residual block data, and U.S. patent application case has all been described the coding method that encoding block indicates for US2005/0169374A1 number and US2005/0169374A1 number.Yet the disclosed preamble reference modelling method of these files not only depends on the information of adjacent block, also depends on the block type of current block.Use the first level (existence of the first level in the 4 * 4block) indications 4 * 4 conversion residual block data of 4 * 4 transform blocks in United States Patent (USP) US7251371B2 number, according to this method, image or frame are split in the predefined fixing piece size.
Summary of the invention
Main purpose of the present invention is exactly at the deficiencies in the prior art, and a kind of method and apparatus that is used for the preamble reference modelling of 4 * 4 block conversion coefficient indications is provided, and can effectively improve the code efficiency of video image and can save required memory space.
For achieving the above object, the present invention is by the following technical solutions:
A kind of method that is used for the preamble reference modelling of 4 * 4 block conversion coefficient indications may further comprise the steps:
A. obtain the binary digit state value of 4 * 4 block end sign by dualization; With
B. the preamble data that will encode turn to the index value of predetermined reference model, and choose described block end sign corresponding reference model according to described index value, described reference model contains the suitable coding parameter that is written into when being used for for described block end identification code.
Preferably, described predetermined reference model has three, and described step b comprises following substep:
B1. find out top, described 4 * 4 sides two transform blocks adjacent as two reference blocks with left;
B2. determine that according to the conversion coefficient of described reference block described index value is as follows:
If two reference blocks all contain the conversion coefficient of non-zero, getting described index value is 2; Another conversion coefficient is zero entirely if one of them reference block has the conversion coefficient of non-zero, and getting described index value is 1; If the conversion coefficient of two reference blocks is zero entirely, getting described index value is 0; With
B3. choose in three reference models one of them according to the index value of determining.
The size of described reference block can be 4 * 4,8 * 8 or 16 * 16.
Described coding parameter comprise at least the most normal appearance symbol (Most Probable Symbol, MPS) and probability (PMPS).
A kind of preamble reference adaptive binary arithmetic coding (CABAC) method of upward using in self adaptation size conversion system (ABT) comprises the step that realizes preamble reference modelling according to aforesaid method, also further may further comprise the steps:
C. according to the coding parameter of previous step gained described block end sign is encoded, produce code stream; With
D. upgrade the coding parameter in the selected reference model.
A kind of device that is used for the preamble reference modelling of 4 * 4 block conversion coefficient indications comprises:
The dualization module is used for the binary digit state value by the block end sign of 4 * 4 of dualization acquisitions; With
The preamble reference modelling module, the preamble data that are used for having encoded turn to the index value of predetermined reference model, and choose described block end sign corresponding reference model according to described index value, described reference model contains the suitable coding parameter that is written into when being used for for described block end identification code.
Preferably, described preamble reference modelling module comprises:
Reference block is searched the unit, is used to find out top, described 4 * 4 sides two transform blocks adjacent with left as two reference blocks;
The index value determining unit is used for determining in the following manner described index value:
If two reference blocks all contain the conversion coefficient of non-zero, getting described index value is 2; Another conversion coefficient is zero entirely if one of them reference block has the conversion coefficient of non-zero, and getting described index value is 1; If the conversion coefficient of two reference blocks is zero entirely, getting described index value is 0; With
Reference model is chosen the unit, be used for according to the index value of determining choose three predetermined reference models one of them.
A kind of preamble reference adaptive binary arithmetic coding (CABAC) that uses in self adaptation size conversion (ABT) system installs, and comprises the device of aforesaid preamble reference modelling, also further comprises:
Block end identification code module is used for according to the coding parameter of previous step gained described block end sign being encoded, and produces code stream; With
The coding parameter update module is used to upgrade the coding parameter in the selected reference model.
Beneficial technical effects of the present invention is:
1, can point out the optimum value of EOB coding parameter according to device of the present invention and method.Owing to compare with global statistics, the character of EOB and partial statistics (not containing current transform block) have more correlation, in the present invention, preamble reference modelling only depends on the preamble data of having encoded, this just can improve the especially forecasting accuracy of probability parameter of coding parameter than conventional method, these parameters is made to drive encoder EOB is encoded, and can effectively improve the efficient of encoding video pictures, reduce the data volume that needs transmission, improve coding efficiency.
2, compare with the H.264 standard that the information that not only depends on adjacent block also depends on the block type of current block, device of the present invention and method only need less preamble reference model, and preamble reference model still less can be saved more memory space.
3, compare with the method for first level that uses 4 * 4 transform blocks, the transform block among the present invention needs not to be the good fixed dimension of predefine, and can be 4 * 4,8 * 8 or 16 * 16.
Description of drawings
Fig. 1 is the basic procedure of CABAC coding method of the present invention;
Fig. 2 is the implementing procedure of the CABAC coding method of the embodiment of the invention, wherein comprises the step of the preamble reference modelling method of the embodiment of the invention;
Fig. 3 is three nonrestrictive examples according to the reference block choosing method of embodiment of the present invention.
Fig. 4 is the basic structure of CABAC code device of the present invention.
The present invention only limits the process of dualization and choosing of preamble reference model, and other partly can be implemented by the ordered method of different coding standard;
Feature of the present invention and advantage will be elaborated in conjunction with the accompanying drawings by embodiment.
Embodiment
Shown in Figure 1 is the basic procedure of CABAC coding method, and it can be divided into following step:
1) dualization, 2) preamble reference modelling, 3) the binary arithmetic coding, 4) the reference model renewal.
1) dualization is exactly will coded data, i.e. the EOB of 4 * 4 fritters turns to and only comprises 0 and 1 binary digit state.EOB represents that with 04 * 4 fritters do not have the conversion coefficient of non-zero, and promptly all coefficients all are zero; Conversion coefficient with the rarest non-zero of 1 expression 4 * 4 fritters.
2) preamble reference modelling turns to preamble coded data the index of predetermined reference model exactly, and the coding parameter in the indexed reference model is only coding parameter.These parameters comprise the symbol of the most normal appearance and probability thereof etc. at least, and these parameters indicate next step binary arithmetic coding institute to go out coding.
3) binary arithmetic coding is encoded according to the parameter and the current coded data of wanting of previous step gained, produces code stream.
4) reference model upgrades, and upgrades the coding parameter in the indexed reference model.
Before the EOB of all pieces coding is finished, repeat above-mentioned steps unit to the last.
The flow process of a kind of embodiment of this method as shown in Figure 2, concrete steps are as follows.
Obtain the value X (step S1) of current transform block 4 * 4EOB.If 4 * 4 current transform blocks do not have the conversion coefficient of non-zero, X is 0; If the conversion coefficient of or above non-zero is arranged, X is 1.
The reference model one of 4 * 4EOB has three, all contains the conditional probability parameter that EOB encodes in three reference models.Each one of them reference model of 4 * 4EOB index uses.The method of the index value of acquisition reference model is as follows:
Find out two adjacent transform blocks as reference piece (step S2), two adjacent reference blocks are respectively the adjacent transform block of top and left, the adjacent transform block above representing with A, and B represents the adjacent transform block of left.A and B reference block size can be arbitrarily, for example 4 * 4,8 * 8 or 16 * 16, determine by the selected transform size of macro block.Three examples shown in Fig. 3 a-3b, adjacent block can be in same macro blocks, also can be respectively in neighboring macro-blocks.Among Fig. 3 a-3b, the dotted line lattice are represented current transform block, and the horizontal line lattice are represented the transform block that left is adjacent, and the ordinate lattice are represented the transform block that the top is adjacent.
The index value of reference model depends on that A and B have the not conversion coefficient (step S3) of non-zero:
If A, the B reference block all has the conversion coefficient of non-zero, and value is 2;
If A, in the B reference block, getting only one has the conversion coefficient of non-zero and another conversion coefficient is zero entirely, and value is 1;
If A, the conversion coefficient of B reference block are zero entirely, value is 0.
After obtaining index value, choose one of them (step S4) in three reference models according to index value,
Read the coding parameter that this reference model comprises, comprise (step S5) such as MPS, PMPS.
Next drive the binary arithmetic coding, according to the coding parameter that is written into to X encode (step S6).
Upgrade the coding parameter (step S7) in the selected reference model behind the encoder output code flow again.
The related partial statistics of above embodiment is only relevant with adjacent transform block, and this comes the prediction probability parameter more accurate than the use global statistics.Thus, this method can provide more accurate preamble reference model, and its accuracy can be weighed by the state of conditional probability, through observation, the level of this probability can be increased to and be higher than 80% from being lower than 70% according to the preamble reference modelling of this method.This method has improved code efficiency, and the slip of the data that pass is reached to 0.27% of every 1bit.
The implementation method that above-mentioned binary arithmetic coding, the structure of reference model and reference model upgrade can be implemented by different coding standards, also can upgrade the parameter of reference model behind the coding by the relevant defined method of standard, and these variations are unrestricted.
On the other hand, also provide a kind of device that is used for the preamble reference modelling of 4 * 4 block conversion coefficient indications, comprising:
The dualization module is used for the binary digit state value by the block end sign of 4 * 4 of dualization acquisitions; With
The preamble reference modelling module, the preamble data that are used for having encoded turn to the index value of predetermined reference model, and choose described block end sign corresponding reference model according to described index value, described reference model contains the suitable coding parameter that is written into when being used for for described block end identification code.
In a preferred embodiment, described preamble reference model module comprises again:
Reference block is searched the unit, is used to find out top, described 4 * 4 sides two transform blocks adjacent with left as two reference blocks;
The index value determining unit is used for determining in the following manner described index value:
If two reference blocks all contain the conversion coefficient of non-zero, getting described index value is 2; Another conversion coefficient is zero entirely if one of them reference block has the conversion coefficient of non-zero, and getting described index value is 1; If the conversion coefficient of two reference blocks is zero entirely, getting described index value is 0; With
Reference model is chosen the unit, be used for according to the index value of determining choose three predetermined reference models one of them.
Aspect another, also provide a kind of preamble reference adaptive binary arithmetic coding (CABAC) device that in self adaptation size conversion (ABT) system, uses.Please refer to Fig. 4, the device that it comprises aforesaid preamble reference modelling also further comprises:
Block end identification code module is used for according to the coding parameter of previous step gained described block end sign being encoded, and produces code stream; With
The coding parameter update module is used to upgrade the coding parameter in the selected reference model.
The various embodiments of aforementioned means all can specifically be implemented with reference to the content of preceding method embodiment, repeats no more herein.
Above content be in conjunction with concrete preferred implementation to further describing that the present invention did, can not assert that concrete enforcement of the present invention is confined to these explanations.For the general technical staff of the technical field of the invention, without departing from the inventive concept of the premise, can also make some simple deduction or replace, all should be considered as belonging to protection scope of the present invention.
Claims (8)
1. a method that is used for the preamble reference modelling of 4 * 4 block conversion coefficient indications is characterized in that, may further comprise the steps:
A. obtain the binary digit state value of 4 * 4 block end sign by dualization; With
B. the preamble data that will encode turn to the index value of predetermined reference model, and choose described block end sign corresponding reference model according to described index value, described reference model contains the suitable coding parameter that is written into when being used for for described block end identification code.
2. the method for claim 1 is characterized in that, described predetermined reference model has three, and described step b comprises following substep:
B1. find out top, described 4 * 4 sides two transform blocks adjacent as two reference blocks with left;
B2. determine that according to the conversion coefficient of described reference block described index value is as follows:
If two reference blocks all contain the conversion coefficient of non-zero, getting described index value is 2; Another conversion coefficient is zero entirely if one of them reference block has the conversion coefficient of non-zero, and getting described index value is 1; If the conversion coefficient of two reference blocks is zero entirely, getting described index value is 0; With
B3. choose in three reference models one of them according to the index value of determining.
3. method as claimed in claim 2 is characterized in that, the size of described reference block is 4 * 4,8 * 8 or 16 * 16.
4. the method for claim 1 is characterized in that, described coding parameter comprises the symbol and the probability thereof of the most normal appearance at least.
5. a preamble reference adaptive binary arithmetic coding method that uses in self adaptation size conversion system comprises the step that realizes preamble reference modelling according to each described method in the claim 1 to 4, also further may further comprise the steps:
C. according to the coding parameter of previous step gained described block end sign is encoded, produce code stream; With
D. upgrade the coding parameter in the selected reference model.
6. a device that is used for the preamble reference modelling of 4 * 4 block conversion coefficient indications is characterized in that, comprising:
The dualization module is used for the binary digit state value by the block end sign of 4 * 4 of dualization acquisitions; With
The preamble reference modelling module, the preamble data that are used for having encoded turn to the index value of predetermined reference model, and choose described block end sign corresponding reference model according to described index value, described reference model contains the suitable coding parameter that is written into when being used for for described block end identification code.
7. device as claimed in claim 6 is characterized in that, described preamble reference model module comprises:
Reference block is searched the unit, is used to find out top, described 4 * 4 sides two transform blocks adjacent with left as two reference blocks;
The index value determining unit is used for determining in the following manner described index value:
If two reference blocks all contain the conversion coefficient of non-zero, getting described index value is 2; Another conversion coefficient is zero entirely if one of them reference block has the conversion coefficient of non-zero, and getting described index value is 1; If the conversion coefficient of two reference blocks is zero entirely, getting described index value is 0; With
Reference model is chosen the unit, be used for according to the index value of determining choose three predetermined reference models one of them.
8. a preamble reference adaptive binary arithmetic coding device that uses in self adaptation size conversion system comprises according to the device of claim 6 or 7 described preamble reference modellings, also further comprises:
Block end identification code module is used for according to the coding parameter of previous step gained described block end sign being encoded, and produces code stream; With
The coding parameter update module is used to upgrade the coding parameter in the selected reference model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 200910168826 CN101646086B (en) | 2009-08-21 | 2009-08-21 | Method and device of preamble reference modelling of 4*4 block conversion coefficient indication |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 200910168826 CN101646086B (en) | 2009-08-21 | 2009-08-21 | Method and device of preamble reference modelling of 4*4 block conversion coefficient indication |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101646086A true CN101646086A (en) | 2010-02-10 |
| CN101646086B CN101646086B (en) | 2011-07-20 |
Family
ID=41657764
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 200910168826 Active CN101646086B (en) | 2009-08-21 | 2009-08-21 | Method and device of preamble reference modelling of 4*4 block conversion coefficient indication |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101646086B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113556560A (en) * | 2010-04-13 | 2021-10-26 | Ge视频压缩有限责任公司 | Coding of significance maps and transform coefficient blocks |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2003043346A1 (en) * | 2001-11-16 | 2003-05-22 | Ntt Docomo, Inc. | Image encoding method, image decoding method, image encoder, image decode, program, computer data signal, and image transmission system |
| CN100403801C (en) * | 2005-09-23 | 2008-07-16 | 联合信源数字音视频技术（北京）有限公司 | Adaptive entropy coding/decoding method based on context |
| US8275045B2 (en) * | 2006-07-12 | 2012-09-25 | Qualcomm Incorporated | Video compression using adaptive variable length codes |
-
2009
- 2009-08-21 CN CN 200910168826 patent/CN101646086B/en active Active
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113556560A (en) * | 2010-04-13 | 2021-10-26 | Ge视频压缩有限责任公司 | Coding of significance maps and transform coefficient blocks |
Also Published As
| Publication number | Publication date |
|---|---|
| CN101646086B (en) | 2011-07-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103430539B (en) | Decoded picture buffer management | |
| CN104185990B (en) | The decoding of coefficient in video coding | |
| CN100488254C (en) | Entropy coding method and decoding method based on text | |
| CN104221375B (en) | Grade during rate-distortion optimisation quantifies determines | |
| CN103299620B (en) | Video coding use most probable scanning sequence the scanning sequence information for video block is carried out efficient coding | |
| CN103220508B (en) | Coding and decoding method and device | |
| CN102783035B (en) | Parallel entropy coding and decoding methods and devices | |
| CN104221378B (en) | It is used for the uniform particle size of quantization matrix in video coding | |
| CN104025603A (en) | Progressive coding of position of last significant coefficient | |
| CN103299628A (en) | Indicating intra-prediction mode selection for video coding using CABAC | |
| CN103238322A (en) | Separately coding the position of a last significant coefficient of a video block in video coding | |
| CN103339938A (en) | Performing motion vector prediction for video coding | |
| CN103329529A (en) | Signaling quantization parameter changes for coded units in high efficiency video coding (HEVC) | |
| CN104969555A (en) | Progressive refinement with temporal scalability support in video coding | |
| US20130188734A1 (en) | High throughput binarization (htb) method for cabac in hevc | |
| CN103609112A (en) | Internal bit depth increase in video coding | |
| CN103703776A (en) | Coefficient scanning in video coding | |
| CN104025457A (en) | Context optimization for last significant coefficient position coding | |
| CN103190147A (en) | Joint coding of syntax elements for video coding | |
| CN102884792A (en) | Methods and apparatus for unified significance map coding | |
| CN103636225A (en) | Signaling syntax elements for transform coefficients for sub-sets of leaf-level coding unit | |
| CN103636220A (en) | Method and apparatus of intra mode coding | |
| CN103988505A (en) | Constrained reference picture sets in wave front parallel processing of video data | |
| CN103227917A (en) | Decoding method and device | |
| CN104904222A (en) | Signalling of picture order count to timing information relations for video timing in video coding |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |