CN101605262B - Method and device for predicting size motion of variable block - Google Patents
Method and device for predicting size motion of variable block Download PDFInfo
- Publication number
- CN101605262B CN101605262B CN 200910100536 CN200910100536A CN101605262B CN 101605262 B CN101605262 B CN 101605262B CN 200910100536 CN200910100536 CN 200910100536 CN 200910100536 A CN200910100536 A CN 200910100536A CN 101605262 B CN101605262 B CN 101605262B
- Authority
- CN
- China
- Prior art keywords
- macro block
- pixel
- block pixels
- current macro
- pixels
- 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.)
- Active
Links
Images
Landscapes
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
The invention provides a method for predicting size motion of a variable block. The method comprises the followings steps of: continuously extracting a current macro block pixel which is used for predicting a candidate motion vector from a frame to be predicted, and a reference macro block pixel corresponding to the current macro block pixel, storing the current macro block pixel and the reference macro block pixel in a current macro block pixel memory and a searching window pixel memory, and repeating the step until all candidate motion vectors in a searching window are traversed; extracting the current macro block pixels for the undersampling and pixel truncation treatment; simultaneously extracting the reference macro block pixels for the undersampling and pixel truncation treatment; working out SAD values of the treated reference macro blocks corresponding to the candidate motion vectors in various macro block dividing modes; and selecting the minimum SAD value in various macro block dividing modes and the motion vector corresponding to the minimum SAD value. The invention adopts the full pipeline structure, and the execution period of the whole motion estimation operation is short; moreover, the invention can also realize the real-time treatment and the balance on performance, power consumption and area of an encoder, and has strong commonality.
Description
Technical field
The present invention relates to image and technical field of video coding in the signal processing, particularly predicting size motion of variable block technology in cataloged procedure.
Background technology
In video compression, in order better to remove the redundancy of video information, reach the purpose that when keeping video quality, improves compression ratio as far as possible, extensively adopt the block-based hybrid coding method of predictive coding and transition coding combination, by transition coding the spatial domain image mapped is arrived frequency coefficient, remove the spatial coherence between pixel effectively, then made full use of temporal correlation between consecutive frame based on the predictive coding of estimation and motion compensation, thereby obtained compression performance preferably.
Because the piece coupling has lower hardware complexity, for actual motion, be considered to one of method in common, therefore in VLSI, extensively adopt, nearly all H.261/H.263 and MPEG1, the physical combination of 2/4 coding-decoding circuit all is used for estimation to the piece coupling.
In order to obtain minimum predicated error, no longer adopt fixed block size pattern when in H.264/AVC, carrying out inter prediction, but adopt the variable block length pattern that the sub-piece of each division is carried out estimation and motion compensation, obtaining best match block, thereby improve compression efficiency.H.264/AVC in the multi-mode estimation, (Macroblock MB) can further be divided into the experimental process piece and compensate one 16 * 16 macro block.The macroblock partition pattern comprises 16 * 16 as shown in Figure 1,16 * 8,8 * 16,8 * 8; For 8 * 8 sub-pieces, can further be divided into 8 * 4,4 * 8,4 * 4 block modes, the concrete application as shown in Figure 2, the optimal prediction modes that the current macro of Fig. 2 searches out in the region of search of reference frame is that upper left 8 * 8 sub-piece optimization models are two 8 * 4 sub-pieces, is Mv0 and Mv1 to deserved motion vector, the optimal prediction modes of its excess-three 8 * 8 sub-piece is 8 * 8, is respectively Mv2, Mv3 and Mv4 to deserved motion vector.The macroblock partition pattern has improved the relevance that reaches the sub-piece of interior macroblocks between each macro block greatly.To travel through all predictive modes when H.264/AVC the macro block of encoding being carried out motion prediction, promptly all to carry out hunting action one time to the sub-piece of each size, therefore inter prediction just can move at the various of different big wisps in the image, utilize optimal sub-piece size to make up, with the correctness of raising search, and can be because of the error that causes with the different big or small objects of the piece removal search of fixed dimension when mobile, producing.Variable block length block matching motion search example in Figure 3 shows that H.264/AVC, in this example, through the multi-mode estimation, current macro is divided into five sub-pieces, the motion vector of the corresponding different directions of each sub-piece.
After determining macroblock partition pattern (size of piece), need calculate the motion vector of each sub-piece correspondence to be encoded to each subblock coding, then each motion vector of coding transmission and macroblock partition pattern.Therefore when a selected bigger piece size, then need less code word to transmissions of encoding of minute block message and motion vector, still inaccessible accurate coupling that is to say that the residual error at image partly may comprise many information; And selected less piece size can significantly reduce the information of residual error part, represents motion vector and corresponding partition strategy but produce than higher code check.Therefore need to select suitable partition strategy under different conditions, the quality of piece partition strategy directly affects the efficient and the compression effectiveness of coded system.
The method that two kinds of general motion predictions are arranged at present, a kind of method is the motion prediction that various macroblock partition patterns is carried out one by one the fixed block size, this method is owing to all need traversal search window zone to macroblock partition pattern in each, required storage operation bandwidth is huge, simultaneously required computing time is long, can't realize real-time processing.Another method is handled based on the sub-piece with minimum macroblock partition size, after calculating the SAD residual values of these boy's pieces, obtain the SAD residual values of bigger sub-piece by the combination of some boy's pieces, this method need be stored more results of intermediate calculations, and required temporary resource is more.
Summary of the invention
The objective of the invention is in order to overcome the deficiency of prior art, proposed predicting size motion of variable block method in a kind of real-time processing H.264/AVC, this method can reduce hardware spending, and required storage operation bandwidth is little, computing time is short, and required temporary resource is few.
The present invention has proposed a kind of predicting size motion of variable block device simultaneously.
The predicting size motion of variable block method comprises the steps:
(1) from external memory storage, continues to take out current macro block pixels and the corresponding reference macro block pixels thereof that is used to predict a certain candidate motion vector in the frame to be predicted, be stored in current macro block pixels memory and search window pixel memories respectively, all candidate motion vectors in the traversal search window;
(2) take out the current macro block pixels of a candidate motion vector from the current macro block pixels memory, and carry out down-sampled and the pixel cut position is handled current macro block pixels, with the result cache handled at the current macro block pixels register array;
When (3) carrying out step (2), take out the reference macroblock pixel of a candidate motion vector from the search window pixel memories, and carry out down-sampled and the pixel cut position is handled to the reference macroblock pixel of taking out, with the result cache handled at reference macroblock pixel register array;
(4) current macro block pixels after step (2) and step (3) are handled and reference macroblock pixel are delivered to the SAD computing unit by current macro block pixels register array and reference macroblock pixel register array respectively and are carried out computing in one-period, calculate the sad value of this candidate motion vector corresponding reference macro block under various macroblock partition patterns;
(5) sad value under the various macroblock partition patterns that obtain of sad value under the various macroblock partition patterns and previous candidate motion vector computing compares, and chooses minimum sad value and corresponding motion vector under the various macroblock partition patterns.
Continuity for the ease of prediction processing between level two candidate motion vectors adjacent with vertical direction, described reference macroblock pixel register array moves up or down the several rows pixel in vertical direction in a computing cycle, or is moved to the left some row pixels in the horizontal direction.
The predicting size motion of variable block device, comprise: the current macro block pixels memory, the search window pixel memories, the down-sampled processing unit one of macro block pixels, the down-sampled processing unit two of macro block pixels, pixel cut position processing unit one, pixel cut position processing unit two, the current macro block pixels register array, reference macroblock pixel register array, SAD computing unit, SAD comparing unit, motion prediction flow process control unit, wherein:
Described current macro block pixels memory continues to take out the current macro block pixels that is used to predict a certain candidate motion vector in the frame to be predicted from external memory storage;
Described search window pixel memories continues to take out current macro block pixels corresponding reference macro block pixels from external memory storage.
The down-sampled processing unit one of macro block pixels, the down-sampled processing unit two of macro block pixels take out current macro block pixels and the reference macroblock pixel is carried out down-sampled processing from current macro block pixels memory and search window pixel memories respectively independently, it reduces current macro block pixels register array and reference macroblock pixel register array under the prerequisite that does not influence described SAD computational accuracy size reduces the hardware resource consumption of SAD computing unit simultaneously;
Pixel cut position processing unit one, pixel cut position processing unit two carry out the processing of pixel cut position to current macro block pixels after the described down-sampled processing and reference macroblock pixel respectively independently, and it reduces the hardware resource consumption of SAD computing unit under the prerequisite that does not influence described SAD computational accuracy;
The current macro block pixels that the current macro block pixels register array is temporary after down-sampled and cut position is handled through described macro block pixels, register array offers described SAD computing unit with the pixel of whole macro block in one-period;
The reference macroblock pixel that reference macroblock pixel register array is temporary after down-sampled and pixel cut position is handled through described macro block pixels, register array offers described SAD computing unit with the pixel of whole macro block in one-period;
Continuity for the ease of the prediction processing between level two candidate motion vectors adjacent with vertical direction, reference macroblock pixel register array can be in the vertical direction pixel of dividing a word with a hyphen at the end of a line that is shifted up or down in a computing cycle, or is moved to the left the structure of some row pixels in the horizontal direction;
The SAD computing unit calculates the similarity between current macro block pixels and the candidate motion vector corresponding reference pixel macroblock, the adding up of the current macro block pixels with described macro block pixels after the down-sampled and cut position processing and the absolute value of the difference on the reference macroblock pixel correspondence position;
The SAD comparing unit is chosen the pairing block mode of minimum sad value under the various macroblock partition patterns and motion vector predictive mode and the motion vector as the current macro position;
The startup of motion prediction flow process control unit control current macro block pixels memory, search window pixel memories, two down-sampled processing units of macro block pixels, two pixel cut position processing units, current macro block pixels register array unit, reference macroblock pixel register array element, macro block SAD computing unit, SAD comparing unit.
Beneficial effect of the present invention is mainly reflected in: this method adopts full flowing structure on the one hand, the whole estimation computing execution cycle is short, required storage operation bandwidth is little, required temporary resource is few, can be implemented between encoder performance, power consumption, the area and weigh, adopt down-sampled and cut position to handle the hardware resource that can under the situation of not decreasing coding efficiency, take still less, very little to the compression efficiency influence of encoder; Because this method is encoded based on block mode, therefore realize and to adopt to have very strong versatility on the other hand in the various video standard.According to the different video coding standard, only need adjust that SAD adds up and mode decision unit or mode selecting unit get final product at macroblock partition size in multiple this standard.
Description of drawings
The various macroblock partition patterns of Fig. 1 for using in the estimation
Fig. 2 is the application example of the various macroblock partition patterns used in the estimation
Fig. 3 is the variable block length block matching motion search example in H.264/AVC
Fig. 4 is the example that concerns of search window zone and hunting zone.
The 1/2 down-sampled template that Fig. 5 adopts for the present invention.
Fig. 6 is the SAD calculating of 8x8 macroblock partition of the present invention and the process of combination
Fig. 7 is predicting size motion of variable block method flow diagram of the present invention and predicting size motion of variable block structure drawing of device
Fig. 8 is not isotopic number pixel cut position result that the embodiment of the invention the realized comparison diagram with the JM8.4 identifying code
Fig. 9 is H.264/AVC encoder RD curve that the embodiment of the invention the realized comparison diagram with the JM8.4 identifying code
Embodiment
Below in conjunction with accompanying drawing content of the present invention is further specified.
The predicting size motion of variable block method comprises the steps, as shown in Figure 7:
(1) from external memory storage, continues to take out current macro block pixels and the corresponding reference macro block pixels thereof that is used to predict a certain candidate motion vector in the frame to be predicted, be stored in current macro block pixels memory and search window pixel memories respectively, all candidate motion vectors in the traversal search window;
(2) take out the current macro block pixels of a candidate motion vector from the current macro block pixels memory, and carry out down-sampled and the pixel cut position is handled current macro block pixels, with the result cache handled at the current macro block pixels register array;
When (3) carrying out step (2), take out the reference macroblock pixel of a candidate motion vector from the search window pixel memories, and carry out down-sampled and the pixel cut position is handled to the reference macroblock pixel of taking out, with the result cache handled at reference macroblock pixel register array;
(4) current macro block pixels after step (2) and step (3) are handled and reference macroblock pixel are delivered to the SAD computing unit by current macro block pixels register array and reference macroblock pixel register array respectively and are carried out computing in one-period, calculate the sad value of this candidate motion vector corresponding reference macro block under various macroblock partition patterns;
(5) sad value under the various macroblock partition patterns that obtain of sad value under the various macroblock partition patterns and previous candidate motion vector computing compares, and chooses minimum sad value and corresponding motion vector under the various macroblock partition patterns.
Continuity for the ease of prediction processing between level two candidate motion vectors adjacent with vertical direction, described reference macroblock pixel register array in a computing cycle vertical direction move up the several rows pixel, move down the several rows pixel in vertical direction, or be moved to the left some row pixels in the horizontal direction.
The predicting size motion of variable block device, as shown in Figure 7, comprise: current macro block pixels memory (1), search window pixel memories (2), the down-sampled processing unit one (3) of macro block pixels, the down-sampled processing unit two (4) of macro block pixels, pixel cut position processing unit one (5), pixel cut position processing unit two (6), current macro block pixels register array (7), reference macroblock pixel register array (8), SAD computing unit (9), SAD comparing unit (10), motion prediction flow process control unit (11), wherein:
Described current macro block pixels memory (1) continues to take out the current macro block pixels of a certain candidate motion vector of prediction in the frame to be predicted from external memory storage;
Described search window pixel memories (2) continues to take out current macro block pixels corresponding reference macro block pixels from external memory storage, the number of the reference pixel macro block of being stored in the search window depends on the hunting zone of being adopted, for example, for horizontal direction [16,15], the hunting zone of vertical direction [16,15], the reference pixel macro block number of required loading are 3x3=9; For horizontal direction [32,31], the hunting zone of vertical direction [32,31], the reference pixel macro block number of required loading is 5x5=25, Figure 4 shows that the number and the hunting zone relation of macro block in the search window;
The down-sampled processing unit one (3) of macro block pixels, the down-sampled processing unit two (4) of macro block pixels take out current macro block pixels and the reference macroblock pixel is carried out down-sampled processing from current macro block pixels memory and search window pixel memories respectively independently, it reduces the size of current macro block pixels register array and reference macroblock pixel register array under the prerequisite that does not influence described SAD computational accuracy, reduce the hardware resource consumption of SAD computing unit simultaneously, Figure 5 shows that a kind of example of down-sampled pattern;
Pixel cut position processing unit one (5), pixel cut position processing unit two (6) carry out the processing of pixel cut position to current macro block pixels after the described down-sampled processing and reference macroblock pixel respectively independently, and it reduces the hardware resource consumption of SAD computing unit under the prerequisite that does not influence described SAD computational accuracy.
The specific implementation that the pixel cut position is handled is the processing that pixel is moved to right, and keeps the high position data after the displacement, handles as the cut position that 8 pixels is kept 5, and only need move to right 3 gets final product.The result of the not isotopic number pixel cut position processing of the present invention's realization relatively as shown in Figure 8, benchmark relatively is the JM8.4 identifying code, and the pixel cut position is handled to go bail for respectively from high to low and stayed 7 bit data, reservation 6 bit data, reservation 5 bit data and keep four kinds of situations of 4 bit data.The result shows that when pixel data kept 5, the overall performance of encoder had decline slightly, and signal to noise ratio reduces about 0.1dB; And pixel data is when keeping 4, and it is more that the overall performance of encoder descends, and signal to noise ratio reduces about 0.3dB.
The current macro block pixels that current macro block pixels register array (7) is temporary after down-sampled and cut position is handled through described macro block pixels, register array offers described SAD computing unit with the pixel of whole macro block in one-period.
The reference macroblock pixel that reference macroblock pixel register array (8) is temporary after down-sampled and pixel cut position is handled through described macro block pixels, register array offers described SAD computing unit with the pixel of whole macro block in one-period.
Continuity for the ease of the prediction processing between level two candidate motion vectors adjacent with vertical direction, reference macroblock pixel register array (8) in a computing cycle can vertical direction move up the several rows pixel, in the vertical direction one-row pixels that is shifted downwards, or be moved to the left the structure of some row pixels in the horizontal direction.
SAD computing unit (9) calculates the similarity between current macro block pixels and the candidate motion vector corresponding reference pixel macroblock, the adding up of the current macro block pixels with described macro block pixels after the down-sampled and cut position processing and the absolute value of the difference on the reference macroblock pixel correspondence position.The process that a kind of SAD by minimum macroblock partition obtains the calculating of SAD of big macroblock partition and combination as shown in Figure 6, this figure is depicted as the SAD calculating of a 8x8 macroblock partition and the process of combination, and the SAD calculating of other macroblock partition and anabolic process are similarly.
SAD comparing unit (10) comprises SAD comparing unit one and SAD comparing unit two,
Wherein, SAD comparing unit one, selected a certain macroblock partition pattern, the sad value of more different candidate motion vectors under this macroblock partition, thus select a motion vector with minimum sad value.
Wherein, SAD comparing unit two, in the result that SAD comparing unit one provides, the sad value of the pairing minimum of more various macroblock partition selects wherein the pairing macroblock partition pattern of reckling and motion vector as the predictive mode and the motion vector of current macro position.
Another realization of SAD comparing unit is to comprise SAD comparing unit three and SAD comparing unit four,
Wherein, SAD comparing unit three, the sad value of more same candidate motion vector under various macroblock partition, thus select a sub-block mode with minimum sad value.
Wherein, SAD comparing unit four, in the result that SAD comparing unit three provides, the sad value of the minimum of each candidate motion vector correspondence relatively selects wherein the pairing block mode of reckling and motion vector as the predictive mode and the motion vector of current macro position.
The startup of motion prediction flow process control unit (11) control current macro block pixels memory, search window pixel memories, two down-sampled processing units of macro block pixels, two pixel cut position processing units, current macro block pixels register array unit, reference macroblock pixel register array element, macro block SAD computing unit, SAD comparing unit.
Be illustrated in figure 9 as H.264/AVC encoder RD curve ratio that the present invention realizes, benchmark is the JM8.4 identifying code.The result shows that the H.264/AVC encoder that the present invention realized is under identical code check, and than the JM8.4 identifying code, signal to noise ratio reduces about 0.1dB.
What should be understood that is, the foregoing description is just to explanation of the present invention, rather than limitation of the present invention, and any innovation and creation that do not exceed in the connotation scope of the present invention all fall within the protection range of the present invention.
Claims (4)
1. the predicting size motion of variable block method is characterized in that comprising the steps:
(1) from external memory storage, continues to take out current macro block pixels and the corresponding reference macro block pixels thereof that is used to predict a certain candidate motion vector in the frame to be predicted, be stored in current macro block pixels memory and search window pixel memories respectively, all candidate motion vectors in the traversal search window;
(2) take out the current macro block pixels of a candidate motion vector from the current macro block pixels memory, and carry out down-sampled and the pixel cut position is handled current macro block pixels, with the result cache handled at the current macro block pixels register array;
When (3) carrying out step (2), take out the reference macroblock pixel of a candidate motion vector from the search window pixel memories, and carry out down-sampled and the pixel cut position is handled to the reference macroblock pixel of taking out, with the result cache handled at the reference pixel register array;
(4) current macro block pixels after step (2) and step (3) are handled and reference macroblock pixel pass out to the SAD computing unit by current macro block pixels register array and reference macroblock pixel register array respectively and carry out computing in one-period, calculate the sad value of this candidate motion vector corresponding reference macro block under various macroblock partition patterns;
(5) sad value under the various macroblock partition patterns that obtain of sad value under the various macroblock partition patterns and previous candidate motion vector computing compares, and chooses minimum sad value and corresponding motion vector under the various macroblock partition patterns.
2. predicting size motion of variable block method according to claim 1, it is characterized in that described reference pixel register array in a computing cycle vertical direction move up the several rows pixel, move down the several rows pixel or be moved to the left some row pixels in the horizontal direction in vertical direction.
3. predicting size motion of variable block device, it is characterized in that comprising: the current macro block pixels memory, the search window pixel memories, the down-sampled processing unit one of macro block pixels, the down-sampled processing unit two of macro block pixels, pixel cut position processing unit one, pixel cut position processing unit two, the current macro block pixels register array, reference macroblock pixel register array, SAD computing unit, SAD comparing unit, motion prediction flow process control unit, wherein:
Described current macro block pixels memory continues to take out the current macro block pixels that is used to predict a certain candidate motion vector in the frame to be predicted from external memory storage;
Described search window pixel memories continues to take out current macro block pixels corresponding reference macro block pixels from external memory storage;
The down-sampled processing unit one of macro block pixels, the down-sampled processing unit two of macro block pixels take out current macro block pixels and the reference macroblock pixel is carried out down-sampled processing from current macro block pixels memory and search window pixel memories respectively independently;
Pixel cut position processing unit one, pixel cut position processing unit two carry out the processing of pixel cut position to current macro block pixels after the described down-sampled processing and reference macroblock pixel respectively independently;
The current macro block pixels that the current macro block pixels register array is temporary after down-sampled and cut position is handled through described macro block pixels, the reference macroblock pixel that reference macroblock pixel register array is temporary after down-sampled and pixel cut position is handled through described macro block pixels, current macro block pixels register array and reference macroblock pixel register array offer described SAD computing unit with the pixel of whole macro block respectively in one-period under the control of motion prediction control unit;
The SAD computing unit calculates the similarity between current macro block pixels and the candidate motion vector corresponding reference pixel macroblock, the adding up of the current macro block pixels with described macro block pixels after the down-sampled and cut position processing and the absolute value of the difference on the reference macroblock pixel correspondence position;
The SAD comparing unit is chosen the pairing block mode of minimum sad value under the various macroblock partition patterns and motion vector predictive mode and the motion vector as the current macro position;
The startup of motion prediction flow process control unit control current macro block pixels memory, search window pixel memories, two down-sampled processing units of macro block pixels, two pixel cut position processing units, current macro block pixels register array unit, reference macroblock pixel register array element, macro block SAD computing unit, SAD comparing unit.
4. as predicting size motion of variable block device as described in the claim 3, it is characterized in that described reference macroblock pixel register array in a computing cycle can vertical direction move up the several rows pixel, move down the several rows pixel or be moved to the left some row pixels in the horizontal direction in vertical direction.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 200910100536 CN101605262B (en) | 2009-07-09 | 2009-07-09 | Method and device for predicting size motion of variable block |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 200910100536 CN101605262B (en) | 2009-07-09 | 2009-07-09 | Method and device for predicting size motion of variable block |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101605262A CN101605262A (en) | 2009-12-16 |
| CN101605262B true CN101605262B (en) | 2011-11-09 |
Family
ID=41470780
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 200910100536 Active CN101605262B (en) | 2009-07-09 | 2009-07-09 | Method and device for predicting size motion of variable block |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101605262B (en) |
Families Citing this family (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR101419634B1 (en) * | 2010-04-07 | 2014-07-16 | 가부시키가이샤 제이브이씨 켄우드 | Moving image encoding device, moving image encoding method, moving image encoding program, moving image decoding device, moving image decoding method, and moving image decoding program |
| KR101456499B1 (en) * | 2010-07-09 | 2014-11-03 | 삼성전자주식회사 | Method and apparatus for encoding and decoding motion vector |
| CN102340617B (en) * | 2010-07-14 | 2014-06-11 | 奇景光电股份有限公司 | Motion estimation (ME) and motion compensation (MC) circuit |
| CN102227132B (en) * | 2011-05-25 | 2017-04-12 | 深圳市云宙多媒体技术有限公司 | Motion vector prediction coding and decoding method, apparatus thereof and coding and decoding system |
| GB2493755B (en) * | 2011-08-17 | 2016-10-19 | Canon Kk | Method and device for encoding a sequence of images and method and device for decoding a sequence of images |
| CN104427345B (en) * | 2013-09-11 | 2019-01-08 | 华为技术有限公司 | Acquisition methods, acquisition device, Video Codec and its method of motion vector |
| CN107087168B (en) * | 2017-03-31 | 2020-07-31 | 武汉斗鱼网络科技有限公司 | Method and device for judging intra-frame prediction mode |
| CN110337810B (en) * | 2018-04-02 | 2022-01-14 | 北京大学 | Method and apparatus for video processing |
| CN112738517B (en) * | 2019-10-14 | 2022-03-01 | 珠海格力电器股份有限公司 | Motion estimation search method, device, equipment and storage medium |
| CN111968151B (en) * | 2020-07-03 | 2022-04-05 | 北京博雅慧视智能技术研究院有限公司 | Motion estimation fine search method and device |
| 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 |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1333634A (en) * | 2001-01-12 | 2002-01-30 | 北京航空航天大学 | Quick video motion estimating method |
| CN1589022A (en) * | 2004-08-26 | 2005-03-02 | 中芯联合（北京）微电子有限公司 | Macroblock split mode selecting method in multiple mode movement estimation decided by oriented tree |
| CN1753501A (en) * | 2005-10-31 | 2006-03-29 | 连展科技(天津)有限公司 | Method of selecting H.264/AVC frame to frame motion estimation mode |
| CN101009833A (en) * | 2006-01-23 | 2007-08-01 | 三星电子株式会社 | Method of and apparatus for deciding encoding mode for variable block size motion estimation |
| CN101404774A (en) * | 2008-11-13 | 2009-04-08 | 四川虹微技术有限公司 | Macro-block partition mode selection method in movement search |
-
2009
- 2009-07-09 CN CN 200910100536 patent/CN101605262B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1333634A (en) * | 2001-01-12 | 2002-01-30 | 北京航空航天大学 | Quick video motion estimating method |
| CN1589022A (en) * | 2004-08-26 | 2005-03-02 | 中芯联合（北京）微电子有限公司 | Macroblock split mode selecting method in multiple mode movement estimation decided by oriented tree |
| CN1753501A (en) * | 2005-10-31 | 2006-03-29 | 连展科技(天津)有限公司 | Method of selecting H.264/AVC frame to frame motion estimation mode |
| CN101009833A (en) * | 2006-01-23 | 2007-08-01 | 三星电子株式会社 | Method of and apparatus for deciding encoding mode for variable block size motion estimation |
| CN101404774A (en) * | 2008-11-13 | 2009-04-08 | 四川虹微技术有限公司 | Macro-block partition mode selection method in movement search |
Also Published As
| Publication number | Publication date |
|---|---|
| CN101605262A (en) | 2009-12-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101605262B (en) | Method and device for predicting size motion of variable block | |
| CN1882094B (en) | Image encoding device, image encoding method, image decoding device, image decoding method, and communication device | |
| CN104967857B (en) | The method and apparatus that image is encoded/decoded | |
| CN101087417B (en) | Picture coding device | |
| CN110225346A (en) | A kind of decoding method and its equipment | |
| CN111200736B (en) | Method and device for acquiring motion information in video coding/decoding | |
| CN101267556B (en) | Quick motion estimation method and video coding and decoding method | |
| CN102291581B (en) | Realizing method of self-adaptive motion estimation supporting frame field | |
| CN102047665A (en) | Dynamic image encoding method and dynamic image decoding method | |
| CN103997645B (en) | Quick HEVC intra-frame coding unit and pattern decision method | |
| CN101686393A (en) | Fast-motion searching method and fast-motion searching device applied to template matching | |
| JP7330243B2 (en) | Method and apparatus for encoding or decoding video data in FRUC mode with reduced memory access | |
| CN103248895A (en) | Quick mode estimation method used for HEVC intra-frame coding | |
| CN108989799B (en) | Method and device for selecting reference frame of coding unit and electronic equipment | |
| CN108134939A (en) | A kind of method for estimating and device | |
| CN102647598B (en) | H.264 inter-frame mode optimization method based on maximin MV (Music Video) difference value | |
| CN103957414A (en) | HEVC intra-frame prediction coding method and system | |
| CN101505427A (en) | Movement estimation apparatus in video compression encoding algorithm | |
| CN105025298B (en) | The method and apparatus that image is encoded/decoded | |
| CN103384327A (en) | AVS fast mode selection algorithm based on adaptive threshold | |
| CN102265615B (en) | Use the image prediction of the subregion again in reference cause and effect district and use the Code And Decode of such prediction | |
| CN102710940A (en) | Digital signal processor (DSP) platform-based method for realizing optimization of video coding by utilizing motion estimation algorithm | |
| CN109688411B (en) | Video coding rate distortion cost estimation method and device | |
| CN201639717U (en) | Size-changeable motion predicting device | |
| JP2011010297A (en) | System and method for estimating sum of absolute differences |
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 |