CN101383971A - Intra-frame prediction processing method based on image encoding and decoding - Google Patents
Intra-frame prediction processing method based on image encoding and decoding Download PDFInfo
- Publication number
- CN101383971A CN101383971A CN 200710121440 CN200710121440A CN101383971A CN 101383971 A CN101383971 A CN 101383971A CN 200710121440 CN200710121440 CN 200710121440 CN 200710121440 A CN200710121440 A CN 200710121440A CN 101383971 A CN101383971 A CN 101383971A
- Authority
- CN
- China
- Prior art keywords
- prediction
- piece
- row
- pixel
- frame
- 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.)
- Pending
Links
Images
Landscapes
- Compression Or Coding Systems Of Tv Signals (AREA)
- Compression, Expansion, Code Conversion, And Decoders (AREA)
Abstract
The invention belongs to the designing field of an integrate circuit with video compression, relates to a processing method for the structure of an intra-frame predicting apparatus and pixel predicting seriation of a macro block or a block, provides a processing module aiming at the intra-frame predicting and gives some examples correspondingly, wherein the method can effectively improve the parallel stream of the intro-frame predicting. The key point of the processing method is as follows: a reference pixel and predicting values needed during the predicting process are separately stored, thereby the reference pixel of the next predicting unit (block or macro block) can be dynamically renewed while the intra-frame predicting is performed, in this way, in one predicting unit, sequences can be flexibly arranged between rows, between lines and between pixels so as to perform the intra-frame predicting and enable the infra-frame predicting based on the pixel or pixel row processing to flexibly arrange the sequences. Blocks can be streamed together to realize more efficient processing, which has significance for high-definition and real-time decoding.
Description
Technical field:
The present invention is the integrated circuit (IC) design field that belongs to video compression, relates to the structure of intra predictor generator and the tactic processing method of pixel prediction of macro block or piece.
Background technology:
In the coding and decoding video design, the unit of handling can be macro block (block of pixels of 16x16), also can be sub-piece (as the block of pixels of 8x8 or 4x4), a macro block can be divided into 4 8x8 pieces and 16 4x4 pieces, 4x4,8x8, three kinds of module units of 16x16 are arranged H.264/AVC, have only the 8x8 module unit among the AVS.
The infra-prediction techniques that adopts in the video compress is in order to remove the spatial redundancies of present image, utilize the neighbor of the top of current block and left as a reference, current block is predicted, so just need to when current block is predicted, the data of neighbor must be obtained.When prediction, need relate to the order of pixel prediction, present way is that the order according to raster scan is predicted in piece, be example pixel prediction sequential process in piece when prediction once commonly used is described with the 8x8 piece below, see shown in Figure 1, when beginning to predict piece 0, reference pixel around it has been decoded and has been obtained, according to order from 1 to 8 delegation of delegation of raster scan from top to bottom from left to right order predict, after finishing, piece 0 prediction carries out the prediction of piece 1, the forecasting sequence of the predict pixel of piece 1 also is the same prediction from top to bottom, the piece 2 of back, piece 3 forecasting sequences all are like this.
This forecasting sequence has certain limitation on hardware is realized, because the data after every prediction decoding finishes during infra-frame prediction need when being next piece prediction a bit, such as 8 pixel values of rightmost one row of piece among Fig. 10 is left side reference pixel sample as it when piece 1 prediction, if forecasting sequence according to routine, the left side reference pixel that the back piece 1 that finishes is rebuild in the 8th row prediction just obtains fully, at this moment just can carry out the prediction of piece 1, be that piece 1 needs certain stand-by period, so generally the implementation method at infra-frame prediction all is to carry out stream treatment with the behavior unit in the piece, though can realize the stream treatment of pixel or pixel column in the piece, but can't carry out the stream treatment of pixel or pixel column between piece and the piece, restrict the speed of infra-frame prediction.
Summary of the invention:
The present invention is directed to the deficiency of above-mentioned conventional Predicting Technique, proposed a kind of new infra-frame prediction structure, as shown in Figure 2, infra-frame prediction is divided into preprocessing part and predicted portions.
Preprocessing process is before the prediction reference pixel of current block top and left to be carried out certain processing in the conducting frame to obtain some intermediate object programs, uses afterwards these intermediate object programs to predict.Preprocessing process can be different according to different standards, can be directly to copy with reference to pixel, also can comprise addition and shifting function, even can also be to contain complicated multiply operation.Decompose so following advantage arranged:
1. but preprocessing process and forecasting process stream treatment can improve the performance of infra-frame prediction;
2. the new pixel that generates during the processing of prediction in the conducting frame can be upgraded reference pixel, is used for the prediction of next predicting unit (piece or macro block), does not also affect predicting the outcome of current block simultaneously.
Use this result just to make the forecasting sequence of current block to adjust very flexibly, for referencial use when whether being the subsequent block prediction according to pixel in the current prediction piece, chronological order flexibly, needs pixel as a reference can be predicted in advance, like this in the process of prediction current block, the reference pixel that subsequent block needs has been rebuild and finished, and is ready for the piece prediction that is right after in advance.This flexibly sequence arrangement can carry out according to concrete needs, can be according to from top to bottom line by line prediction in every, from left to right by column prediction, from right to left by column prediction, also can predict line by line from top to bottom and from top to bottom, converge toward the centre, also can be from left to right and prediction etc. various ways by column from right to left.
Because the order of prediction can be arranged flexibly, so according to the different realizations forecasting sequence in the arranged blocks flexibly.So both can arrange flexibly to carry out the stream treatment take row or column as unit in the frame, improve intra-prediction process efficient (stream treatment can not paused during the processing block border), also can conveniently cooperate with other algoritic modules (such as IDCT), realize more high efficiency combination.
The correlation that The present invention be directed to pixel between the piece piece in the forecasting process is selected forecasting sequence, the latent period when eliminating like this next piece prediction, the handling capacity of raising concurrency and whole streamline.
Description of drawings:
Fig. 1 is that a macroblock partitions becomes conventional forecasting sequence schematic diagram when predicting in 4 8x8 piece time frames;
Fig. 2 is an infra-frame prediction overall structure schematic diagram of the present invention;
Fig. 3 is a kind of ranks forecasting sequence of 8x8 piece schematic diagram during the AVS infra-frame prediction among the present invention;
Fig. 4 is 16x16 ranks forecasting sequence schematic diagram during infra-frame prediction H.264 among the present invention;
Fig. 5 is 4x4 ranks forecasting sequence schematic diagram during infra-frame prediction H.264 among the present invention;
Embodiment:
Describe the implementation structure of a kind of infra-frame prediction of the present invention's proposition below in detail:
Two registers are set respectively in concrete implementation procedure (also can use RAM, together lower), a prediction needed reference pixel register of current block and a renewal reference pixel register, be used for leaving the process prediction of piece prediction in and rebuild complete as the required reference pixel of subsequent block prediction, just can the pixel value of follow-up needs be deposited in when prediction and upgrade in the pixel register by register is set, can not have influence on the prediction of current block, can freely carry out very flexibly the arrangement of pixel prediction order, in the prediction current block, can obtain in advance the reference data of next piece.
A macro block can be divided into 4 8x8 pieces, 1 8x8 piece can be divided into 4 4x4 pieces, respectively in conjunction with AVS and intraframe prediction algorithm H.264, concrete example illustrates several forecasting sequences below, can see that structure according to the invention can effectively improve infra-frame prediction efficient.
Embodiment 1:
The concrete forecasting sequence of AVS8x8 piece is seen Fig. 2, the arithmetic element of preprocessing process is according to the difference of the intra prediction mode of AVS and difference, when being level or vertical mode, do not require calculation, directly data are copied to the prediction desirable value register from the reference pixel value register; When for other pattern of brightness, reference pixel need to be through being deposited into after adder and the shift unit computing in the prediction desirable value register; When being colorimetric prediction plane predictive mode, pretreated result was deposited in the prediction desirable value register after reference pixel need to pass through adder, shift unit and complicated multiplier computing.Because the rightmost of piece 0 row need to be as the reference pixel of piece 1, therefore when prediction piece 0, predict by column from right to left with the unit of classifying as, it is complete that on the right of like this when prediction the 3rd row are listed as reconstruct, for piece 1 prediction is got ready, and the reference pixel when last pixel of every row is piece 2 prediction, also can be in the process of piece 1 prediction reconstruct complete, for piece 2 predictions are got ready; Because need the pixel of following delegation of piece 1 for referencial use during piece 2 prediction, can when 1 prediction of prediction piece, adopt from the bottom up line by line prediction, be convenient to like this current drainage water between each piece, nethermost delegation predicts that at first also reconstruct is complete, top reference pixel as piece 2 and piece 3 predictions, last pixel of every delegation is the reference pixel of next macro block then, also can obtain when prediction piece 2; Adopt from right to left by column prediction during piece 2 prediction, prediction is for current drainage water equally like this, and rightmost one is listed as the complete left side predicted value as piece 3 of prediction reconstruct that can do sth. in advance like this, and each is listed as last pixel as the reference pixel of following macro block equally; Just use from right to left by column prediction during piece 3 prediction, predict that at first the pixel that reconstruct is come out is the left side reference pixel of next macro block, last pixel of each row is the top reference pixel of following macro block, prediction processing unit flowing water in each piece can be got up like this, prediction module is being carried out work always in each cycle, can not cause picture wait space between piece and piece during conventional forecasting sequence, serious reduction the running frequency of streamline.
Describe below implementation step in detail:
Step 1: predict first row from right to left according to the adjacent pixel values of the piece 0 that has obtained;
Step 2: the secondary series of prediction piece 0, simultaneously reconstruct first row;
Step 3: the 3rd row of prediction piece 0, reconstruct secondary series simultaneously, first last pixel that is listed as that first row that reconstruct is finished and reconstruct finish saves as piece 1 and piece 2 is prepared;
Step 4: the 4th row of prediction piece 0, the 3rd row of reconstruct simultaneously, last pixel of the secondary series that reconstruct is finished saves as piece 2 and prepares;
Prediction to piece 0 subsequent treatment unit and piece 2 and piece 3 repeats above-mentioned steps, the difference of piece 1 is from the bottom up line by line prediction of prediction, after the complete the first row of reconstruct data are preserved, after complete subsequent rows reconstruct is complete last data of every row are preserved, as the left side reference pixel of next macroblock prediction.
Embodiment 2:
H.264 the interior schematic diagram of predicting take the 16x16 macro block as predicting unit of frame is seen Fig. 3, two registers are set before prediction, one is the REG1 of the required reference pixel of Storage Estimation, REGO for storage current macro prediction and the reconstruct rightmost one row pixel after complete, the reference during as next macroblock prediction of rightmost one row pixel among the REGO, the arithmetic element of preprocessing process is according to the difference of H.264 intra prediction mode and difference, when being level or vertical mode, do not require calculation, directly data are copied to the prediction desirable value register from the reference pixel value register; When for other pattern of brightness, reference pixel need to be through being deposited into after adder and the shift unit computing in the prediction desirable value register; When being colorimetric prediction plane predictive mode, the preliminary treatment result that reference pixel need to pass through after adder, shift unit and the complicated multiplier computing is deposited in the prediction desirable value register.
Step 1: before prediction, obtain earlier current macro required above and the reference pixel of left be stored among the REG1;
Step 2: begin to predict prediction first row according to reference pixel according to from right to left order by row;
Step 3: this moment, the first row prediction was complete, entered reconfiguration unit, began simultaneously the prediction of secondary series;
Step 4: first is listed as, and reconstruct finishes, and according to being saved among the REGO, the left side reference of so next macro block obtains with this columns, and the secondary series of reconstruct is simultaneously predicted the 3rd row;
The step of each follow-up row is the same with above-mentioned three row, until the 16th row prediction is complete.
Embodiment 3:
H.264 the forecasting sequence schematic diagram of middle 4x4 piece is seen Fig. 4, the pretreatment unit of preprocessing process is the same with the structure of above-mentioned 16x16, according to the difference of predictive mode and difference, 4 4x4 pieces in first 8x8 piece are as example, and the forecasting sequence of its excess-three 8x8 piece is the same with first.
Step 1: during first 4x4 piece prediction, adopt from right to left by column prediction, row of at first prediction reconstruction in this piece and last value of every row are preserved;
Step 2: during second 4x4 piece prediction, adopt from the bottom up line by line prediction, doing like this is to be convenient to piece and piece current drainage water, and delegation and last value of every row that at first prediction in this piece is rebuild are preserved;
Step 3: when the 3rd 4x4 piece predicted, adopting from right to left by column prediction, in like manner is convenient for current drainage water, and row of at first prediction reconstruction in this piece and last value of every row are preserved;
Step 4: when the 4th piece 4x4 predicts, adopting from right to left by column prediction, is convenient for flowing water equally, and row of at first prediction reconstruction in this piece and last value of every row are preserved;
4 4x4 piece predictions in such 8x8 piece are complete, and the prediction of the 4x4 piece in three follow-up 8x8 pieces repeats above-mentioned steps.
Claims (6)
1. processing method based on the infra-frame prediction of image coding and decoding, employing is separately preserved the pretreated result who uses in reference pixel and the prediction, carry out based on the preliminary treatment result during infra-frame prediction, the separately storage of predicted value that reference pixel and forecasting process are required, can dynamically update the reference pixel of next predicting unit (piece or macro block) when making prediction in the conducting frame, like this at a predicting unit expert and row, can put in order flexibly between row and row and pixel and the pixel and carry out infra-frame prediction, but the processing sequence flexible dispatching satisfies the needs of different implementation algorithms.
2. the processing method of a kind of infra-frame prediction based on image coding and decoding as claimed in claim 1, it is characterized in that preprocessing process can be different according to different standards, can be directly reference pixel to be copied, also can comprise addition and shifting function, even can also be to contain complicated multiply operation, rebuild pixel during the processing of prediction in the conducting frame like this and can upgrade reference pixel, be used for the prediction of next predicting unit (piece or macro block), also do not influence predicting the outcome of current block simultaneously.
3. the processing method of a kind of infra-frame prediction based on image coding and decoding as claimed in claim 1, it is characterized in that being provided with two registers in the process of various prediction, one is used for preserving the whole pixel of the needed reference of prediction, one is used to preserve the preliminary treatment end value, and separate storage can be convenient to the flowing water layout when carrying out the piece prediction.
4. the processing method of a kind of infra-frame prediction based on image coding and decoding as claimed in claim 1, it is characterized in that macro block at 16x16, when prediction, from macro block from right to left 16 row begin prediction by row, after finishing, the first row prediction reconstruct is saved in the reference pixel buffer memory, because prediction is used is value in the predicted value buffer memory, can not have influence on the prediction of prediction subsequent column like this, last value preservation that will be listed as when subsequent column has been rebuild in prediction simultaneously.
5. the processing method of a kind of infra-frame prediction based on image coding and decoding as claimed in claim 1, it is characterized in that the piece for 8x8, a macro block can be divided into 4 8x8 pieces, piece is predicted according to "the" shape during prediction, when predicting first, forecasting sequence is to predict by column from right to left, toward line by line prediction during second prediction from lower, to predict by column from right to left during the 3rd prediction, be to predict by column from right to left during the 4th prediction, after each piece prediction reconstruct is complete, row (OK) of at first reconstruct and last value of subsequent column (OK) be saved in the reference pixel buffer memory.
6. the processing method of a kind of infra-frame prediction based on image coding and decoding as claimed in claim 1, it is characterized in that piece at 4x4, a macro block is divided into 4 8x8 pieces, each 8x8 piece is divided into 4 4x4 pieces, the prediction of 4x4 piece is " it " font prediction in each 8x8 piece in the macro block, and then according to " it " Word prediction 8x8 piece, in each 8x8 piece, the forecasting sequence of 4x4 piece is the same, when first 4x4 piece of prediction, forecasting sequence is from right to left by the row prediction, during second 4x4 piece prediction is to predict line by line from the bottom up, be from right to left by the row prediction during the 3rd 4x4 piece piece prediction, be from right to left by the row prediction during the 4th 4x4 piece piece prediction, after each piece prediction reconstruct finishes, row (OK) of reconstruct at first and last value of subsequent column (OK) are saved in the reference pixel buffer memory, can influence the prediction of piece;
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 200710121440 CN101383971A (en) | 2007-09-06 | 2007-09-06 | Intra-frame prediction processing method based on image encoding and decoding |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 200710121440 CN101383971A (en) | 2007-09-06 | 2007-09-06 | Intra-frame prediction processing method based on image encoding and decoding |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN101383971A true CN101383971A (en) | 2009-03-11 |
Family
ID=40463549
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 200710121440 Pending CN101383971A (en) | 2007-09-06 | 2007-09-06 | Intra-frame prediction processing method based on image encoding and decoding |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101383971A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101894154A (en) * | 2010-07-13 | 2010-11-24 | 重庆大学 | Method for extracting key pattern from image file |
| CN102855197A (en) * | 2011-11-08 | 2013-01-02 | 东南大学 | Storage system implementing method for large-scale coarse-grained reconfigurable system |
| CN103379324A (en) * | 2012-04-16 | 2013-10-30 | 乐金电子(中国)研究开发中心有限公司 | Parallel realization method, device and system for advanced motion vector prediction AMVP |
| CN107079167A (en) * | 2014-10-01 | 2017-08-18 | 高通股份有限公司 | Pipelining infra-frame prediction hardware structure for video coding |
| CN108111856A (en) * | 2010-12-23 | 2018-06-01 | 英国广播公司 | The compression of image |
| CN116074533A (en) * | 2023-04-06 | 2023-05-05 | 湖南国科微电子股份有限公司 | Motion vector prediction method, system, electronic device and storage medium |
| WO2023193701A1 (en) * | 2022-04-06 | 2023-10-12 | 展讯通信（上海）有限公司 | Image coding method and apparatus |
-
2007
- 2007-09-06 CN CN 200710121440 patent/CN101383971A/en active Pending
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101894154A (en) * | 2010-07-13 | 2010-11-24 | 重庆大学 | Method for extracting key pattern from image file |
| CN101894154B (en) * | 2010-07-13 | 2012-02-15 | 重庆大学 | Method for extracting key pattern from image file |
| CN108111856A (en) * | 2010-12-23 | 2018-06-01 | 英国广播公司 | The compression of image |
| CN102855197A (en) * | 2011-11-08 | 2013-01-02 | 东南大学 | Storage system implementing method for large-scale coarse-grained reconfigurable system |
| CN103379324A (en) * | 2012-04-16 | 2013-10-30 | 乐金电子(中国)研究开发中心有限公司 | Parallel realization method, device and system for advanced motion vector prediction AMVP |
| CN103379324B (en) * | 2012-04-16 | 2017-12-15 | 乐金电子(中国)研究开发中心有限公司 | A kind of advanced motion-vector prediction AMVP Parallel Implementation method, apparatus and system |
| CN107079167A (en) * | 2014-10-01 | 2017-08-18 | 高通股份有限公司 | Pipelining infra-frame prediction hardware structure for video coding |
| CN107079167B (en) * | 2014-10-01 | 2020-07-17 | 高通股份有限公司 | Method, apparatus and system for performing intra prediction |
| WO2023193701A1 (en) * | 2022-04-06 | 2023-10-12 | 展讯通信（上海）有限公司 | Image coding method and apparatus |
| CN116074533A (en) * | 2023-04-06 | 2023-05-05 | 湖南国科微电子股份有限公司 | Motion vector prediction method, system, electronic device and storage medium |
| CN116074533B (en) * | 2023-04-06 | 2023-08-22 | 湖南国科微电子股份有限公司 | Motion vector prediction method, system, electronic device and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102547296B (en) | Motion estimation accelerating circuit and motion estimation method as well as loop filtering accelerating circuit | |
| TWI445411B (en) | Method for performing local motion vector derivation during video coding of a coding unit, and associated apparatus | |
| Khan et al. | Hardware-software collaborative complexity reduction scheme for the emerging HEVC intra encoder | |
| CN101383971A (en) | Intra-frame prediction processing method based on image encoding and decoding | |
| CN100586180C (en) | Be used to carry out the method and system of de-blocking filter | |
| CN106973297A (en) | Method for video coding and hybrid video coders | |
| CN106464894A (en) | Method and apparatus for processing video | |
| CN105245897A (en) | Efficient software for transcoding to hevc on multi-core processors | |
| CN103517069A (en) | HEVC intra-frame prediction quick mode selection method based on texture analysis | |
| CN101115207B (en) | Method and device for implementing interframe forecast based on relativity between future positions | |
| CN103414895A (en) | Encoder intra-frame prediction device and method applicable to HEVC standards | |
| CN101222646B (en) | Infra-frame prediction device and method suitable for AVS encoding | |
| CN102238383B (en) | For the multibus system architecture of Video Codec | |
| Jiang et al. | Block-Based Parallel Intra Prediction Scheme for HEVC. | |
| CN103957414A (en) | HEVC intra-frame prediction coding method and system | |
| KR101292668B1 (en) | Video encoding apparatus and method based-on multi-processor | |
| CN102055981A (en) | Deblocking filter for video coder and implementation method thereof | |
| CN102148990B (en) | Device and method for predicting motion vector | |
| Xiao et al. | A fine-grained parallel implementation of a H. 264/AVC encoder on a 167-processor computational platform | |
| CN105791829A (en) | HEVC parallel intra-frame prediction method based on multi-core platform | |
| CN102196272B (en) | P frame encoding method and device | |
| CN1589028B (en) | Predicting device and method based on pixel flowing frame | |
| CN104521234A (en) | Method and apparatus for video processing incorporating deblocking and sample adaptive offset | |
| CN102595137B (en) | Fast mode judging device and method based on image pixel block row/column pipelining | |
| Peesapati et al. | Design of streaming deblocking filter for HEVC decoder |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20090311 |