CN100348047C - Front block and reference block matching computing method and motioning estimating method - Google Patents
Front block and reference block matching computing method and motioning estimating method Download PDFInfo
- Publication number
- CN100348047C CN100348047C CNB2004100966780A CN200410096678A CN100348047C CN 100348047 C CN100348047 C CN 100348047C CN B2004100966780 A CNB2004100966780 A CN B2004100966780A CN 200410096678 A CN200410096678 A CN 200410096678A CN 100348047 C CN100348047 C CN 100348047C
- Authority
- CN
- China
- Prior art keywords
- data
- current block
- memory
- point
- block
- 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 present invention discloses a method for matching and operating the current block and a reference block so as to realize that a processor matches and operates data of each point in the current block and data of each point of the reference block. The present invention comprises that (1) the processor moves the data of each point of the current block stored in an off-hip memory into an on-chip memory in advance; (2) the processor moves the data of each point of the reference block stored in a discontinuous memory address in the off-chip memory to the on-chip memory, and the memory address of the data in the on-chip memory is continued; (3) the data of each point of the current block and the data of each point of a reference point are matched and operated in sequence, and an operation value is obtained. The processor can use a DMA mode for shifting the data of each point in the reference block to the present invention so as to avoid an offset that each point of the reference block calculates the point and a first address of a reference window. Thereby, a matched speed rate is improved, and further, the processing efficiency of the processor is improved. In addition, the present invention also provides a motion estimating method comprising the mapping course so as to improve the processing efficiency of the processor.
Description
Technical field
The present invention relates to the Digital Video Processing field, relate in particular to a kind of current block and reference block and carry out the method for matching operation and adopt said method to carry out the method for estimation.
Background technology
Now, motion estimation techniques has been widely used in fields such as video compression coding and video format conversion, in order to reduce the needed bandwidth of digital of digital video data of the given picture quality of transmission as far as possible, multiple different video compression algorithm has been developed and has been used for compressed video data.The several multimedia technology standard committee set up and propose to be used to encode/standard of compressed audio and video data.
Method for video coding mainly comprises compression and interframe compression in the frame, and wherein the interframe compression is the main source of its compression ratio, therefore existing video compression standard (MPEG1/2/4, H.261/3 wait), and employing is based on the interframe compression scheme of estimation mostly.The principle of its method for estimating is earlier present frame to be divided into some big or small same blocks (described is referred to as current block), then each current block is set the piece that (described window is referred to as reference window) search is mated with it most in the big or small window in reference frame.The alternate position spike of current block and match block is called motion vector, and the pixel difference is called residual block.Because the pixel near 0 in the residual block is a lot, by processing such as dct transform, quantifications, can increase substantially compression ratio, and then eliminate the time redundancy between frame and the frame.
See also Fig. 1, it is the principle flow chart of method for estimating, and described method for estimating finds the piece that mates most in order to realize current block from reference window, may further comprise the steps:
At first carry out step S110: processor (as DSP) obtains the data of each pixel of current block in the present frame, and determines the hunting zone of reference window in the searching algorithm that adopts and the reference frame;
Carry out step S120 then: processor is determined reference block according to described searching algorithm from reference window;
Carry out step S130 subsequently: processor carries out computing with the data of described each pixel of current block and the data of each respective pixel of reference block according to matching criterior, obtains operation values;
Carry out step S140 subsequently: judge in described reference window, whether to have traveled through all reference blocks according to described searching algorithm, if, then carry out step S150, otherwise, carry out step S120;
Carry out step S150 at last: from all operation values, find minimum value, then with the pairing reference block of described minimum operation values as the piece that mates most with current block.
Obviously, determine several (such as Q) reference blocks according to searching algorithm in reference window, then processor needs current block to carry out matching operation with Q reference block respectively.And, above-mentioned steps only finds piece that current block mates most in the present frame, and the number of current block is more than one in the present frame, that is to say that the treating capacity of the processor of handling estimation is very big, and the processing time is longer when causing processor to carry out estimation simultaneously.
See also Fig. 2, its method for carrying out matching operation between current block in the prior art and the reference block.It comprises:
At first carry out step S210: processor calculates the memory address at each pixel place of reference block and the side-play amount of reference window first address respectively, and described reference window first address is the initial address that processor can be visited the reference window data;
Carry out step S220 subsequently: processor takes out the data of each pixel in the described reference block according to described side-play amount from memory space, and respectively with current block in the data of corresponding pixel calculate according to matching criterior, and then preserve calculated value.
Below be example with the brightness value that obtains in the reference block, come to carry out between labor current block and the reference block process of matching operation.
Suppose that current block formed (seeing also Fig. 3) by 3*3 pixel, reference window is made up of 5*5 pixel, and the piece that the pixel of putting on 1～9 (wherein 1～9 is the identifier of described pixel) among the figure is formed is the reference block (seeing also Fig. 4) of this matching operation, and the current block pixel identical among Fig. 3 and Fig. 4 with identifier in the reference block, corresponding mutually each other pixel.In addition, suppose that the searching algorithm that adopts is full-search algorithm (FS), matching criterior is absolute error criterion (SAD).
The brightness value of each pixel of current block is b
k(m, n), b
k(m, n) being expressed as abscissa is m, ordinate is the brightness value of the pixel of n, m=1～3 wherein, n=1～3.The brightness value of each pixel of reference window is b
K-1(p, q), b
K-1(p, q) being expressed as abscissa is p, ordinate is the brightness value of the pixel of q, p=1～5 wherein, q=1～5.
Reference window with Fig. 4 is an example, and the brightness value of each pixel is a sequential storage in the reference window in memory, as storage [1 successively, 1], [1,2], [1,3], [1,4], [1,5], [2,1], [2,2], [2,3] ... [5,4], [5,5] brightness value, wherein [m, n] expression abscissa is m, and ordinate is the pixel of n.But address stored is discontinuous in each pixel brightness value memory of reference block.Therefore, if processor wants to obtain the brightness value of 9 pixels of reference block from memory, just need calculate the side-play amount offset k (k is 1～9) of the memory address and the initial address (also be first address) of 5*5 reference window in memory of these 9 pixels earlier; According to described side-play amount, again the brightness value of each pixel correspondence is read from memory respectively then.
According to P[k]=5* (m-1)+(n-1) calculates the side-play amount of each reference point, wherein, P[k] be the side-play amount of k pixel and reference window first address, m is the abscissa at k pixel place, n is the ordinate at k pixel place.
Obviously, need to calculate the side-play amount offset of 9 pixels, processor will carry out the computing of 9 multiplication and 9 sub-additions.And the number of pixels of forming current block is many more, and the operation times that processor need carry out is also many more, thereby causes processor efficient in coupling computational process not high, and the efficient that coupling is calculated is not high yet.
Adopt above-mentioned current block and reference block to carry out the method for estimating of matching operation, because current block need all need carry out matching operation with each reference block that searching algorithm described in the reference window is determined, and need find more than one of the number of the current block of match block by matching operation, therefore to mate the number of times of calculating more for processor, thereby cause the treatment effeciency of processor low, and then influence the speed of whole estimation, influenced the processing speeds such as compression of image thus.
Summary of the invention
The object of the present invention is to provide a kind of current block and reference block to carry out the method for matching operation, to solve in the prior art because each pixel of the reference block memory address in memory in the reference window is discontinuous, thereby cause processor need calculate each pixel, and then the technical problem of the efficient of the speed of influence coupling calculating and processor processing for the side-play amount of reference window at the first address of memory.
The present invention also provides a kind of method that adopts above-mentioned matching process to carry out estimation.
For addressing the above problem, the invention discloses the method that a kind of current block and reference block carry out matching operation, to realize that processor carries out matching operation with the data of each pixel in the current block and the data of each point of reference block, comprising: (1) processor will be stored in the data in advance of each pixel of the current block of chip external memory and move to on-chip memory; (2) processor will be stored in each point of reference block in the discontinuous memory address in the chip external memory data-moving to on-chip memory, make that the memory address of described data in on-chip memory is continuous; (3) data of described each pixel of current block and the data of each point of reference block are carried out matching operation successively, and then obtain operation values.
Processor utilize dma mode with the data-moving of the data of each point of described reference block and/or each point of current block to described on-chip memory.
Memory capacity in order to the stored reference blocks of data in the on-chip memory can be allocated in advance according to the memory capacity that current block takies.
Described processor is DSP.
The invention also discloses a kind of method for estimating, comprising: (1) processor will be stored in the data in advance of each pixel of current block of chip external memory and move to on-chip memory, and the hunting zone of reference window in searching algorithm that determine to adopt and the reference frame; (2) processor is determined reference block according to described searching algorithm from reference window; (3) processor carries out computing with the data of described each pixel of current block and the data of each respective pixel of reference block according to matching criterior, obtain operation values, specifically comprise: (31) processor will be stored in the data-moving of each pixel of reference block in the discontinuous memory address in the chip external memory to on-chip memory, make that the memory address of described data in on-chip memory is continuous; (32) data of described each pixel of current block and the data of each pixel of reference point are carried out matching operation successively, and then obtain operation values; (4) judge in described reference window, whether to have traveled through all reference blocks according to described searching algorithm, if, carry out step (5), otherwise, carry out step (2); (5) from all operation values, find minimum value, then with the pairing reference block of described minimum operation values as the piece that mates most with current block.
Processor utilize dma mode with the data-moving of each pixel in the data of each pixel in the described reference block and/or the current block to described on-chip memory.
Memory capacity in order to the stored reference blocks of data in the on-chip memory can be allocated in advance according to the memory capacity that current block takies.
Described searching algorithm comprises full-search algorithm, diamond search algorithm and two dimension to searching algorithm, and described matching criterior comprises the absolute error criterion.
Described data comprise the brightness value of pixel.
Compared with prior art, the present invention has the following advantages:
When the present invention carries out matching operation with the data of each point in the data of each point of current block and the reference block, a moving on-chip memory at first moved the data of each point in the reference block by processor from a discontinuous memory space, make that the address of storage is continuous, and then carry out matching operation, avoided each point of reference block to calculate the side-play amount of this point and reference window first address, improve the speed of coupling thus, and then improved the treatment effeciency of processor.And the present invention moves data to on-chip memory from chip external memory in advance, and CPU is significantly smaller than the speed that the time for reading of chip external memory data is improved thus processor processes data to the time for reading of on-chip memory data.
In addition, the present invention also adopts the dma mode moving data, can carry out consistency operation by the dma mode moving data, reduces processor consuming time in when coupling thus, further improves processor processing efficient.
When above-mentioned matching operation process is useful in the method for estimating, because current block needs and a more than reference block carries out mapping operations in the method for estimating, so when matching efficiency improves, the treatment effeciency of whole method for estimating will be largely increased.And, in carrying out the piece matching process, CPU can only carry out matching operation, and the data carrying can be finished on the backstage by dma controller, that is to say when each CPU carries out matching operation that dma controller is carried the data of next reference block to on-chip memory, avoided CPU should computing to carry the step of data again thus, CPU only need carry out matching operation, improves the treatment effeciency of CPU thus greatly.
Description of drawings
Fig. 1 is the principle flow chart of method for estimating;
Fig. 2 is for carrying out the flow chart of the method for matching operation between current block in the prior art and the reference block;
Fig. 3 is 3*3 the current block that pixel is formed;
Fig. 4 is 5*5 the reference window that pixel is formed;
Fig. 5 carries out the flow chart of the method for matching operation for current block of the present invention and reference block;
Fig. 6 is the flow chart of a kind of method for estimating of the present invention.
Embodiment
Below in conjunction with accompanying drawing, specify the present invention.
See also Fig. 5, it carries out the flow chart of the method for matching operation for current block of the present invention and reference block.Described method is carried out matching operation in order to realize processor with the data of each point of reference block in the data of each point (being pixel) in the current block and the reference window, comprising:
At first carry out step S310: processor will be stored in the data in advance of each point of the current block of chip external memory and move to on-chip memory;
Carry out step S320 then: processor will be stored in the data-moving of each point of reference block in the discontinuous memory address in the chip external memory to on-chip memory, make that the memory address of described data in on-chip memory is continuous; Memory capacity in order to the stored reference blocks of data in the on-chip memory can be allocated in advance according to the memory capacity that current block takies.On-chip memory generally, the memory capacity in order to the stored reference blocks of data in the on-chip memory is pre-assigned, and answers on-chip memory to be equal to or greater than to equal the memory capacity that current block takies.Also have, in moving process, processor should consider that the storage characteristics of each point of reference block move successively.Data such as each point of this reference block lay respectively at 3 sections continuous memory spaces in memory, then processor can be docile and obedient preface obtains described reference block respectively from these 3 sections continuous memory spaces each point and be stored in data in the memory space;
Carry out step S320 then: the data of each point of described current block and the data of each point of reference point are carried out matching operation successively, and then obtain operation values.
In said method, because the data-moving of each point of reference block is to on-chip memory, make that each storage is continuous in the memory address of on-chip memory, therefore when carrying out matching operation, need not to calculate the side-play amount of each pixel for the first address of reference window, directly the data of the data of each point of current block and each point of reference block being carried out matching operation successively gets final product, saved the spent time of address arithmetic thus, and then improved the treatment effeciency of processor, also improved the efficient of matching operation simultaneously.
Inventor's processor in step S310 utilize dma mode with the data-moving of each pixel in the described current block to described on-chip memory.Processor can also utilize dma mode with the data-moving of each pixel in the reference block to described on-chip memory.DMA moves and can carry out consistency operation, reduces data-moving consuming time on processor thus.
The matching process of above-mentioned reference block and current block goes for a lot of places, so long as in memory cell, obtain the data of the discontinuous storage of part, and to described data successively with reference block in the data of each point contrast, operation such as calculating, and then finish a certain task, can adopt the matching process of above-mentioned reference block and current block.
Below to carry out method for estimating with the matching process that adopts current block and reference block be example, specify the above-mentioned current block and the matching process of reference block.See also Fig. 5, it is a kind of flow chart of method for estimating.It may further comprise the steps:
Step S410: processor will be stored in the data in advance of each point of the current block of chip external memory and move to on-chip memory, and and determine the hunting zone of reference window in the searching algorithm that adopts and the reference frame.The searching algorithm that the present invention adopts can comprise that full search method, diamond search algorithm and two dimension are to being applied in all searching algorithms in the estimation in the prior aries such as searching algorithm;
Step S420: processor is determined reference block according to described searching algorithm from reference window;
Step S430: processor carries out computing with the data of described each pixel of current block and the data of each respective pixel of reference block according to matching criterior, obtains operation values, specifically comprises:
S510: processor will be stored in the discontinuous memory address in the chip external memory reference block each
The data-moving of point makes that to on-chip memory the memory address of described data in on-chip memory is continuous;
S520: the data of each point of described current block and the data of each point of reference point are carried out matching operation successively, and then obtain operation values.Processor utilize dma mode with the data-moving of each pixel in the described reference block to described on-chip memory;
S440: judge in described reference window, whether to have traveled through all reference blocks according to described searching algorithm, if, carry out step S450, otherwise, carry out step S420;
S450: from all operation values, find minimum value, then with the pairing reference block of described minimum operation values as the piece that mates most with current block.
Step S510 adopt dma controller by dma mode with the data-moving of each point of reference block to described on-chip memory.When CPU carried out matching operation among the step S520, dma controller was carried the datamation of each point of next reference block among the step S510, to improve the treatment effeciency of CPU.Promptly in carrying out the piece matching process, CPU can only carry out matching operation, and data carryings can be finished on the backstage by dma controller, has avoided CPU should computing to carry the step of data again thus, CPU only need carry out matching operation, improves the treatment effeciency of CPU thus greatly.
Below be example to adopt full search block matched motion method of estimation, concrete method for estimating is described.
At first, determine the size of reference window, promptly at the (M+2d of reference frame according to the full-search algorithm that is adopted
M) * (N+2d
M) the hunting zone in find the best matching blocks of current block, M wherein, N is respectively the level of current block and vertical pixel count, d
MLevel and Vertical number of pixels for the translation of current block maximum possible;
Then, processor is searched for the piece that " mates most " with current block in the reference window of determining, the matching criterior that present embodiment adopted is absolute error criterion (SAD), promptly
B in the following formula
k(m, n) and b
K-1(m, ((i j) is side-play amount, and span is [d for m, brightness value n) n) to represent present frame and reference frame pixel respectively
M, d
M], M, N are the level and the Vertical number of pixels of sub-piece, d
MLevel and Vertical number of pixels for the maximum possible translation.When left side value SAD (i j) hour, represents two sub-pieces couplings, this moment (i j) is considered to the shift value of current block and match block.
According to searching algorithm earlier with sub-piece corresponding in the reference window with current block as the reference piece, carry out matching operation, move along level and vertical direction individual element then, different sub-pieces is set as the reference piece, carry out the matching operation with current block, promptly whenever move and once need calculate SAD one time, then all SAD are compared, finding minimum SAD, right (i j) is the shift value of current block and match block.
Promptly at (M+2d
M) * (N+2d
M) the hunting zone in, total mobile number of times is (2d
M+1)
2, and according to existing matching process, the calculation times that processor need carry out all side-play amounts is (2d
M+ 1)
2* r, r is total number of the pixel that current block comprised.And adopting method of estimation of the present invention, the side-play amount that need not above-mentioned number of times is calculated, and improves the processing speed of processor thus.
For each traversal point (current block and a reference block mate calculating), prior art is carried out M*N multiplication and M*N sub-addition than the present invention more.d
MLevel and Vertical number of pixels for the maximum possible translation.The reference window size is (M+2d so
M) * (N+2d
M).Then full-search algorithm traversal point number is (2d
m+ 1)
2Promptly under full search condition, prior art is carried out (2d than the present invention more
m+ 1)
2* M*N multiplication and (2d
m+ 1)
2* M*N sub-addition.As can be seen from the above, the present invention has improved the efficient of processor when carrying out estimation greatly.
Next be example still, be about to each pixel brightness value of reference block in the 5*5 reference window, carry out matching operation, and obtain the piece of the coupling of described current block in view of the above with the brightness value of each pixel of current block with Fig. 4 and Fig. 3.
At first processor is by in the serialgram built-in storage of dma mode with data-moving to a 3*3 of each pixel of current block, and the first address of establishing this connected storage is address2.Because address space is continuous, each point carried out the computing of side-play amount needn't be again in prior art, and expect the brightness value b of reference block
K-1(p q), directly reads from the memory block and gets final product.
We relatively use the matching process of matching process of the present invention and prior art, and for each traversal point (being that current block and a reference block carry out matching operation), prior art is carried out 9 multiplication and 9 sub-additions than the present invention more.If adopt full-search algorithm, then a traversal point number is 9, and promptly prior art is carried out 9*9=81 multiplication and 9*9=81 sub-addition than the present invention more.If the DSP of processor adopting C6204 chip, then each multiplying order 2 clock cycle consuming time, each addition instruction 1 clock cycle consuming time, then prior art is than the present invention 2*81+81=243 how consuming time clock cycle.
More than disclosed only be several specific embodiment of the present invention, but those skilled in the art can think variation, all should drop in protection scope of the present invention, those skilled in the art can think variation, all should drop in protection scope of the present invention.
Claims (10)
1, a kind of current block and reference block carry out the method for matching operation, to realize that processor carries out matching operation with the data of each pixel in the current block and the data of each point of reference block, it is characterized in that, comprising:
(1) the processor data in advance of each pixel that will be stored in the current block of chip external memory is moved to on-chip memory;
(2) processor will be stored in each point of reference block in the discontinuous memory address in the chip external memory data-moving to on-chip memory, make that the memory address of described data in on-chip memory is continuous;
(3) data of described each pixel of current block and the data of each point of reference block are carried out matching operation successively, and then obtain operation values.
2, current block as claimed in claim 1 and reference block carry out the method for matching operation, it is characterized in that, adopt in step (1) and the step (2) dma controller by dma mode with the data-moving of the data of each point of described reference block and/or each point of current block to described on-chip memory, step (3) is carried out matching operation by CPU.
3, current block as claimed in claim 1 or 2 and reference block carry out the method for matching operation, it is characterized in that, the memory capacity in order to the stored reference blocks of data in the on-chip memory can be allocated in advance according to the memory capacity that current block takies.
4, current block as claimed in claim 1 or 2 and reference block carry out the method for matching operation, it is characterized in that, described processor is DSP.
5, a kind of method for estimating is characterized in that, comprising:
(1) the processor data in advance that will be stored in each pixel of current block of chip external memory is moved to on-chip memory, and determines the hunting zone of reference window in the searching algorithm that adopts and the reference frame;
(2) processor is determined reference block according to described searching algorithm from reference window;
(3) data of described each pixel of current block and the data of each respective pixel of reference block are carried out computing according to matching criterior, obtain operation values, specifically comprise:
(31) processor will be stored in each pixel of reference block in the discontinuous memory address in the chip external memory data-moving to on-chip memory, make that the memory address of described data in on-chip memory is continuous;
(32) data of described each pixel of current block and the data of each pixel of reference point are carried out matching operation successively, and then obtain operation values;
(4) judge in described reference window, whether to have traveled through all reference blocks according to described searching algorithm, if, carry out step (5), otherwise, carry out step (2);
(5) from all operation values, find minimum value, then with the pairing reference block of described minimum operation values as the piece that mates most with current block.
6, method for estimating as claimed in claim 5 is characterized in that,
Adopt in the step (1) dma controller by dma mode with the data-moving of each point of current block to on-chip memory;
Step (31) adopt dma controller by dma mode with the data-moving of each point of reference block to described on-chip memory.
7, method for estimating as claimed in claim 6 is characterized in that, when CPU carried out matching operation in the step (32), step (31) dma controller was carried the datamation of each point of next reference block, so that improve the treatment effeciency of CPU.
As claim 5 or 7 described method for estimating, it is characterized in that 8, the memory capacity in order to the stored reference blocks of data in the on-chip memory can be allocated in advance according to the memory capacity that current block takies.
As claim 5 or 7 described method for estimating, it is characterized in that 9, described searching algorithm comprises full-search algorithm, diamond search algorithm and two dimension to searching algorithm, described matching criterior comprises the absolute error criterion.
10, as claim 5 or 7 described method for estimating, it is characterized in that described data comprise the brightness value of pixel.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2004100966780A CN100348047C (en) | 2004-12-03 | 2004-12-03 | Front block and reference block matching computing method and motioning estimating method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2004100966780A CN100348047C (en) | 2004-12-03 | 2004-12-03 | Front block and reference block matching computing method and motioning estimating method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1615023A CN1615023A (en) | 2005-05-11 |
| CN100348047C true CN100348047C (en) | 2007-11-07 |
Family
ID=34766524
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2004100966780A Active CN100348047C (en) | 2004-12-03 | 2004-12-03 | Front block and reference block matching computing method and motioning estimating method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100348047C (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN100384216C (en) * | 2004-12-31 | 2008-04-23 | 北京中星微电子有限公司 | Video data compression method and apparatus with optimized searching algorithm |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20030081675A1 (en) * | 2001-10-29 | 2003-05-01 | Sadeh Yaron M. | Method and apparatus for motion estimation in a sequence of digital images |
| JP2004096310A (en) * | 2002-08-30 | 2004-03-25 | Fujitsu Ltd | Program broadcasting apparatus, method and program therefor |
| US6757330B1 (en) * | 2000-06-01 | 2004-06-29 | Hewlett-Packard Development Company, L.P. | Efficient implementation of half-pixel motion prediction |
-
2004
- 2004-12-03 CN CNB2004100966780A patent/CN100348047C/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6757330B1 (en) * | 2000-06-01 | 2004-06-29 | Hewlett-Packard Development Company, L.P. | Efficient implementation of half-pixel motion prediction |
| US20030081675A1 (en) * | 2001-10-29 | 2003-05-01 | Sadeh Yaron M. | Method and apparatus for motion estimation in a sequence of digital images |
| JP2004096310A (en) * | 2002-08-30 | 2004-03-25 | Fujitsu Ltd | Program broadcasting apparatus, method and program therefor |
Also Published As
| Publication number | Publication date |
|---|---|
| CN1615023A (en) | 2005-05-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN1270543C (en) | Method and device for computing coding dynamic images at fixed complicacy | |
| US7940844B2 (en) | Video encoding and decoding techniques | |
| CN1144472C (en) | Motion compensation encoder, decoder and motion compensation encoding and decoding method | |
| CN1525762A (en) | A coding/decoding end bothway prediction method for video coding | |
| CN1316856A (en) | Exercise estimator | |
| CN104067619A (en) | Video decoder, video encoder, video decoding method, and video encoding method | |
| CN1144469C (en) | Motion-vector coding method | |
| CN1713731A (en) | Apparatus and method for estimating hybrid block-based motion | |
| CN1823328A (en) | Motion estimation with fast search block matching | |
| CN1627825A (en) | Motion estimation method for motion picture encoding | |
| US20040008779A1 (en) | Techniques for video encoding and decoding | |
| CN1852442A (en) | Layering motion estimation method and super farge scale integrated circuit | |
| CN1193621C (en) | Approximate IDCT for scalable video and image decoding of computational complexity | |
| CN1112654C (en) | Image processor | |
| CN110392265A (en) | Inter frame motion estimation method, apparatus, electronic equipment and readable storage medium storing program for executing | |
| CN100348047C (en) | Front block and reference block matching computing method and motioning estimating method | |
| CN105100799A (en) | Method for reducing intraframe coding time delay in HEVC encoder | |
| CN1520187A (en) | System and method for video data compression | |
| CN103327340B (en) | A kind of integer searches method and device | |
| CN1449197A (en) | B image mode determining method and apparatus of video coding system | |
| CN1201589C (en) | Motion estimation | |
| CN1535026A (en) | Method of determing reference image block under direct coding mode | |
| CN110662040B (en) | Depth map CU (CU) rapid partitioning method based on reconfigurable array processor | |
| CN1655618A (en) | Search method for video frequency encoding based on motion vector prediction | |
| CN100496126C (en) | Image coder and coding method |
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 |