CN102075748B - Search method for video coding motion estimation - Google Patents
Search method for video coding motion estimation Download PDFInfo
- Publication number
- CN102075748B CN102075748B CN 201010609255 CN201010609255A CN102075748B CN 102075748 B CN102075748 B CN 102075748B CN 201010609255 CN201010609255 CN 201010609255 CN 201010609255 A CN201010609255 A CN 201010609255A CN 102075748 B CN102075748 B CN 102075748B
- Authority
- CN
- China
- Prior art keywords
- absolute difference
- search
- minimum
- point
- enter
- 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.)
- Expired - Fee Related
Links
Images
Abstract
The invention discloses a search method for video coding motion estimation. The search method is implemented by the following steps of: firstly, determining an initial search point by using forecast motion vectors of adjacent blocks of an air domain and a time domain; secondly, selecting different search strategies according to directions of motion vectors of a video sequence, and directly stopping searching static blocks and quasi-static blocks; and finally, adaptively selecting a diamond mod and a triangular mode during searching according to the content characteristic and the motion complexity of the video sequence so as to ensure the search precision and increase the search speed.
Description
Technical field
The present invention relates to a kind of searching method for the Video coding estimation, relate in particular to a kind of needs for the information source coding, based on rhombus and the adaptive Video coding motion estimation searching method of triangle, belong to the digital image processing techniques field.
Background technology
Estimation is to extract the technological package of movable information from video sequence.It is amount of calculation the best part in whole Video coding solution, and in a video compression computing, estimation accounts for 50%~60% of the amount of calculation usually.Therefore, the quality of motion estimation performance directly has influence on the operational efficiency of whole Video coding solution and the reconstruction quality of whole video sequence, finds simply, efficiently, method for estimating is vital fast.
In various method for estimating, block matching method has simply, efficient advantage, therefore by a lot of international video encoding standard, is adopted, and is widely used.Method for estimating based on the piece coupling is the effective solution that reduces time redundancy information.
On the other hand, searching method for estimation has a variety of at present, full searching moving method of estimation (Full Search method wherein, be abbreviated as FS) advantage be the residual error coefficient minimum produced, but its huge time overhead and amount of calculation are that real-time video Coding Project institute is unacceptable.Simultaneously, although FS can search the motion vector that precision is the highest, its motion vector field may not be even, thereby cause overall code efficiency not necessarily optimum.
In order to reduce the motion estimation search complexity of FS method, many improved method for estimating have appearred now, two dimensional logarithmic searching method (TwoDimensional Logarithmic Search as the Jain proposition, TDLS), follow the tracks of the minimum distortion direction and reduced computation complexity.Another kind of famous method is three step searching methods (Three Step Search, TSS), and it only searches for 25 points in the search window, and FS will search for 225 points.There were afterwards a lot of methods to make further modification for TSS, new three step searching methods (New ThreeStep Search, NTSS) for example, the method has been carried out the center-biased mode treatment in the first step.Another kind is effective three step searching methods (Efficient Three Step Search, ETSS), and the method has been applied little diamond pattern in the first step, and does not limit search step in region of search.The people such as Puri have proposed to mix the Orthogonal Search method (OrthogonalSearch, OS) of TDLS and TSS.Intersection search method (Cross Search, CS) and TDLS that Ghanbari proposes are quite similar.The center-biased property that the people such as Po have applied the real video sequence has proposed four-step search method (Four-step Search, 4SS).Within 2000, the people such as Zhu have proposed diamond search method (Diamond Search, DS), and search precision and search speed that the method is remarkable with it are widely used.
An exemplifying embodiment of DS searching method as shown in Figure 1.It has two different fixed modes, and one is large diamond search pattern LDSP (Large Diamond Search Pattern), another little diamond search pattern SDSP (Small Diamond Search Pattern).This searching method can very accurately find overall smallest point.Yet, use fixing search pattern in the processing procedure of DS searching method always, do not consider content character and the motion complexity of video sequence, will search for the point that some there is no need, thereby cause the waste of computational resource.
Summary of the invention
For the deficiencies in the prior art, technical problem to be solved by this invention is to provide a kind of new Video coding motion estimation searching method.This searching method, for the needs of information source coding, carries out the Video coding estimation based on rhombus and triangle self adaptation.
For realizing above-mentioned goal of the invention, the present invention adopts following technical scheme:
A kind of searching method for the Video coding estimation is characterized in that comprising following step:
The 1st step: if the absolute difference of initial search point O and be less than predetermined threshold, search finishes, otherwise enters diamond search pattern; First calculate rhombus four summit A, B, C and D absolute difference and; If the absolute difference of initial search point O and minimum, search finishes; If the summit A that absolute difference and minimum point are this rhombus, enter the 2nd step;
The 2nd step: relatively the absolute difference of B, C, D and O and; The absolute difference of ordering as B or D and be minimum value, enter the 6th step; Otherwise enter the first triangle search pattern, this first triangle search pattern is that the summit A of take in the 1st step is unfaithful intention, take the isosceles triangle that X, Y and Z be summit; Calculate respectively X, Y and Z absolute difference and and with absolute difference and the comparison of A, as absolute difference and minimum that A is ordered, search finishes; As absolute difference and the minimum that X is ordered, enter the 3rd step; As absolute difference and the minimum that Y is ordered, enter the 4th step; As absolute difference and the minimum that Z is ordered, enter the 5th step;
The 3rd step: enter the second triangle search pattern, this second triangle search pattern is to take summit X as unfaithful intention, take the isosceles triangle that B, Y ' and Z ' be summit; Calculate respectively Y ', Z ' absolute difference and and with absolute difference and the comparison of X, as absolute difference and the minimum that X order, search end; As absolute difference and the minimum of Y ' point, return to the 3rd step; Absolute difference and minimum as Z ' point, enter the 4th step;
The 4th step: enter the 3rd triangle search pattern, the 3rd triangle search pattern is to take summit Y as unfaithful intention, take the isosceles triangle that X ', Y ' and Z ' be summit; Calculate respectively X ', Y ', Z ' absolute difference and and with absolute difference and the comparison of Y, as absolute difference and the minimum that Y order, search end; As absolute difference and the minimum of Y ' point, return to the 4th step; Absolute difference and minimum as Z ' point, enter the 5th step; Absolute difference and minimum as X ' point, enter the 3rd step;
The 5th step: enter the 4th triangle search pattern, the 4th triangle search pattern is to take summit Z as unfaithful intention, take the isosceles triangle that D, X ' and Y be summit; Calculate respectively X ', Y ' absolute difference and and with absolute difference and the comparison of Z, as absolute difference and the minimum that Z order, search end; As absolute difference and the minimum of Y ' point, return to the 5th step; Absolute difference and minimum as X ' point, enter the 4th step;
The 6th step: enter the 5th triangle search pattern, the 5th triangle search pattern is to take the isosceles triangle that A, X and Y be summit; Calculate respectively X, Y absolute difference and and with absolute difference and the comparison of A, as absolute difference and minimum that A is ordered, search finishes; Absolute difference and minimum as X or Y are ordered, enter the 7th step;
The 7th step: enter the 6th triangle search pattern, in the situation that the absolute difference that X is ordered and minimum, the 6th triangle search pattern is to take the isosceles triangle that X, A ' and B ' be summit; Calculate respectively A ', B ' absolute difference and and with absolute difference and the comparison of X, as absolute difference and the minimum that X order, search end; Otherwise repeat the 7th step; In the situation that the absolute difference that Y is ordered and minimum the rest may be inferred.
Wherein, in the 1st step, present encoding piece upper left corner coordinate is added to the coordinate position of motion vectors in reference frame of last encoding block is defined as initial search point O.The intermediate value that the motion vectors of encoding block is the adjacent block motion vector, wherein adjacent block comprises three spatial domain adjacent blocks of encoding block described in time domain adjacent block in reference frame and present frame.
Described diamond search pattern is centered by initial search point O, the step-length rhombus that to be 1 pixel form to left and right, upper and lower expansion.
Described threshold value, for detection of going out static block and accurate static block, is directly ended search to static block and accurate static block.
Provided by the present inventionly based on rhombus and the adaptive Video coding motion estimation searching method of triangle, can select adaptively initial search point local minimum to avoid sinking into, searched for for the content character of video sequence and the adaptive selection diamond pattern of motion complexity and triangle pattern in search procedure, improved greatly the search speed of estimation under the prerequisite that guarantees picture quality.
The accompanying drawing explanation
Below in conjunction with the drawings and specific embodiments, the present invention is described in further detail.
Fig. 1 is the search example figure of existing DS searching method;
Fig. 2 is for determining the schematic diagram of initial search point by motion vectors;
Fig. 3 (a)~Fig. 3 (d) is respectively the motion vectors difference profile figure of " Beijing Weather Girl ", " WomanDrinking ", " Circus " and " Mobile Hands " video sequence;
Fig. 4 (a)~Fig. 4 (f) is the search strategy schematic diagram based on rhombus and the adaptive Video coding motion estimation searching method of triangle provided by the present invention;
The search point comparative graph that Fig. 5 (a) is " Beijing Weather Girl " video sequence, the comparative graph of the Y-PSNR that Fig. 5 (b) is " Beijing Weather Girl " video sequence;
The search point comparative graph that Fig. 6 (a) is " Woman Drinking " video sequence, the comparative graph of the Y-PSNR that Fig. 6 (b) is " Woman Drinking " video sequence;
The search point comparative graph that Fig. 7 (a) is " Circus " video sequence, the comparative graph of the Y-PSNR that Fig. 7 (b) is " Circus " video sequence;
The search point comparative graph that Fig. 8 (a) is " Mobile Hands " video sequence, the comparative graph of the Y-PSNR that Fig. 8 (b) is " Mobile Hands " video sequence.
Embodiment
Present patent application has proposed a kind of based on rhombus and the adaptive Video coding motion estimation search of triangle (referred to as DTAS) method.The basic ideas of this DTAS method are: at first utilize the motion vectors of spatial domain and time domain adjacent block to determine initial search point, then the different search strategy according to the set direction of video sequence motion vector, static block and accurate static block are directly ended to search, content character and motion complexity for video sequence in search procedure are selected diamond pattern and triangle pattern adaptively, thereby have effectively improved search speed when guaranteeing search precision.
With reference to figure 2, at first by motion vectors, determine initial search point O.In order to obtain one as far as possible near the initial search point of overall smallest point, select suitable motion vectors to seem very important.In order to obtain the accurate motion vectors of current block, need to consider 2 factors: 1) comprise determining of adjacent block zone (ROS).2) calculate the method for motion vectors.
Comprising adjacent block zone (ROS) is defined as follows: ROS={MB0, and MB1, MB2, MB3}, wherein MB0 means time domain adjacent block in reference frame, and other 3 piece MBi mean the spatial domain adjacent block of encoding block in present frame.If the motion vector of adjacent block is MV0, MV1, MV2 and MV3, the motion vectors MV of encoding block so
currentintermediate value for the adjacent block motion vector.Be defined as follows: MV
current=median (MV0, MV1, MV2, MV3)
Present encoding piece upper left corner coordinate adds the motion vectors MV of last encoding block
currentcoordinate position in reference frame is defined as initial search point O.
The purpose of adaptive motion estimation is exactly by selecting suitable initial search point effectively to adapt to mobile object in region of search, and the benefit of doing like this is easily find real suitable motion vector and reduce amount of calculation.
In this DTAS method, adopt absolute difference and be (SAD) piece coupling distortion function.SAD is defined as:
f wherein
k(m, n) means that the current encoded frame meta is set to the gray value of (m, n), f
k-1(m, n) means that the reference frame meta is set to the gray value of (m, n).(i, j) means the displacement between current encoded image piece and reference image block.
On the other hand, the inventor carries out motion vectors difference profile figure experiment with four video sequences, Fig. 3 (a)~Fig. 3 (d) is respectively the motion vectors difference profile figure of " Beijing WeatherGirl ", " WomanDrinking ", " Circus " and " Mobile Hands " video sequence, and these video sequences have comprised various complexity motion objects.From the distribution map of these motion vectors difference, in the common high concentration of the motion vectors difference zone wide in 2 pixels centered by initial point.Therefore threshold value T is set for detection of going out static block and accurate static block, static block and accurate static block are directly ended to search, thereby accelerate search speed.In this DTAS method, experimental result shows false alarm probability minimum when T is set to 512.
The concrete search strategy of this DTAS method, as shown in Fig. 4 (a)~Fig. 4 (f), comprises following step:
The 1st step: with reference to figure 4 (a), if the absolute difference of initial search point O and SAD (O)<threshold value T, search finishes.Otherwise enter diamond search pattern ABCD, this diamond search pattern is centered by initial search point O, the step-length rhombus that to be 1 pixel form to left and right, upper and lower expansion, four summits that A, B, C and D point are this rhombus.First calculate the sad value of this rhombus four summit A, B, C and D.If the SAD minimum that initial search point O is ordered, search finishes; If one of four summits that minimum SAD point is this rhombus (as the A point, other 3 the like), the 2nd step entered.
The 2nd step: with reference to figure 4 (a), compare the sad value of B, C, D and O each point.The sad value of ordering as B or D is minimum value, enters the 6th step; Otherwise enter triangle search pattern XYZ, this triangle search pattern is that the rhombus summit A of take in the 1st step is unfaithful intention, of formation be take the isosceles triangle XYZ that X, Y and Z are summit.Calculate the sad value of X, Y, Z and compare with the sad value that A is ordered, as the sad value minimum that A is ordered, search finishes; As the sad value minimum that X is ordered, enter the 3rd step; As the sad value minimum that Y is ordered, enter the 4th step; As the sad value minimum that Z is ordered, enter the 5th step.
The 3rd step: with reference to figure 4 (b), enter triangle search pattern BY ' Z ', this triangle search pattern is to take triangular apex X as unfaithful intention, and of formation be take the isosceles triangle BY ' Z ' that B, Y ' and Z ' are summit.Calculate the sad value of Y ', Z ' and compare with the sad value that X is ordered, as the sad value minimum that X is ordered, search finishes; As the sad value minimum of Y ' point, return to the 3rd step; Sad value minimum as Z ' point, enter the 4th step;
The 4th step: with reference to figure 4 (c), enter triangle search pattern X ' Y ' Z ', this triangle search pattern is to take triangular apex Y as unfaithful intention, and of formation be take the isosceles triangle X ' Y ' Z ' that X ', Y ' and Z ' are summit.Calculate the sad value of X ', Y ', Z ' and compare with the sad value that Y is ordered, as the sad value minimum that Y is ordered, search finishes; As the sad value minimum of Y ' point, return to the 4th step; Sad value minimum as Z ' point, enter the 5th step; Sad value minimum as X ' point, enter the 3rd step;
The 5th step: with reference to figure 4 (d), enter triangle search pattern DX ' Y ', this triangle search pattern is to take triangular apex Z as unfaithful intention, and of formation be take the isosceles triangle DX ' Y that D, X ' and Y are summit.Calculate the sad value of X ', Y ' and compare with the sad value that Z is ordered, as the sad value minimum that Z is ordered, search finishes; As the sad value minimum of Y ' point, return to the 5th step; Sad value minimum as X ' point, enter the 4th step;
The 6th step: with reference to figure 4 (e), enter triangle search pattern AXY, this triangle search pattern is that to take rhombus summit A be summit, and of formation be take the isosceles triangle AXY that A, X and Y are summit.Calculate the sad value of X, Y and compare with the sad value that A is ordered, as the sad value minimum that A is ordered, search finishes; As minimum as the sad value that X or Y are ordered (the sad value minimum of ordering as X), enter the 7th step;
The 7th step: with reference to figure 4 (f), enter triangle search pattern XA ' B ', this triangle search pattern is to take triangular apex X as summit, and of formation be take the isosceles triangle XA ' B ' that X, A ' and B ' are summit.Calculate the sad value of A ', B ' and compare with the sad value that X is ordered, as the sad value minimum that X is ordered, search finishes; Otherwise repeat the 7th step.
Above-mentioned DTAS method can solve some defects that exist in the Classical Motion method of estimation, for example the optimal match point of search easily sinks into local smallest point, ignore content character and the motion complexity of video sequence, thereby search for some points that there is no need, cause waste of computational resource etc.
In addition, aspect motion vectors difference rule, the inventor is by a large amount of experiment discoveries, and the distribution of motion vectors difference is high concentration.Utilize the Predicting Technique of motion vector, with the motion vector of the piece of correspondence position in the left side of present encoding piece in present frame, upper, upper right adjacent block and reference frame as motion vectors, can obtain the motion vectors difference profile.
The inventor, by the motion vectors difference profile map analysis to a large amount of dissimilar video sequences, is known the motion complexity height correlation of motion vectors difference profile with video content.For the video of low complex degree, the motion vectors difference is more concentrated to initial search point.This is because level and smooth orderly motion vectors has closer correlation with real motion vector, and produces predicated error still less.Based on this relation, this DTAS method can be by the conversion of diamond pattern and triangle pattern, the adaptively modifying region of search, thus avoid unnecessary search.
The technique effect that this DTAS method realizes can be embodied by the form of curve chart and form.Wherein, compare FS, DS and DTAS method from picture quality and computation complexity two aspects, by Y-PSNR PSNR presentation video quality, count and mean computation complexity by every frame average search, compare the capability and performance of DS and DTAS method with Δ PSNR, with speed-up ratio SUR, describe the speed-raising of DTAS with respect to the DS method.SUR herein is defined as follows: SUR=PN
dS/ PN
dTAS.PN wherein
dSthe search point that means each piece of DS method, PN
dTASthe search point that means each piece of DTAS method.
The inventor is tested with " Beijing Weather Girl ", " Woman Drinking ", these four video sequences of " Circus " and " Mobile Hands ", and these video sequences have comprised various complexity motion objects.The size of piece is fixed to 16 * 16, and hunting zone is ± 7 pixels, only luminance signal is carried out to estimation.
According to Y-PSNR PSNR and the search point of every frame, DS, FS and tri-kinds of methods of DTAS are carried out to performance comparison result as shown in Fig. 5 to Fig. 8, can find out that DTAS method picture quality extremely approaches DS and FS method, and search point significantly descends.
Average peak signal to noise ratio and search point for these four video sequence 100 frames are described as in table 1 and table 2.Table 1 has provided the whole bag of tricks average search and has counted, and by more known, the amount of calculation of DS method is 2.8~10.24 times of DTAS method amount of calculation.Table 2 has provided the average peak signal to noise ratio PSNR of the whole bag of tricks, can find out that the PSNR of DTAS method approaches DS or FS method in ten minutes.Can find out that according to speed-up ratio SUR and Y-PSNR PSNR DTAS method performance, considerably beyond the DS method, reach almost consistent precision, and the amount of calculation of DTAS method is far less than the DS method for dissimilar video sequence.It should be noted that the performance that the DTAS method is showed for dissimilar video sequence all shows suitable stability simultaneously.
Beijing Weather Girl | Woman Drinking | Circus | Mobile Hands | |
FS | 225 | 225 | 225 | 225 |
DS | 12.50 | 17.00 | 16.46 | 22.82 |
DTAS | 1.22 | 5.30 | 5.88 | 5.10 |
SUR | 10.24 | 3.21 | 2.80 | 4.47 |
The comparison (inferior/piece) of counting of table 1 average search
Beijing Weather Girl | Woman Drinking | Circus | Mobile Hands | |
FS | 40.137 | 36.655 | 36.163 | 37.724 |
DS | 40.082 | 36.645 | 36.107 | 37.684 |
DTAS | 39.944 | 36.619 | 36.078 | 37.729 |
ΔPSNR | -0.138 | -0.026 | -0.029 | +0.045 |
Table 2 average peak signal to noise ratio PSNR is (unit: dB) relatively
Above-mentioned experimental result shows, video sequence for various different motion degree, DTAS method provided by the present invention not only can obtain and the essentially identical picture quality of DS method, and speed-up ratio SUR reaches 2.80~10.24, greatly reduce the amount of calculation of piece coupling, so this DTAS method is more applicable for the occasion that requirement of real-time is higher.
The above has been described in detail the searching method for the Video coding estimation of the present invention; for the those skilled in the art of the art, or else deviate from the various apparent change in the situation of the spirit of technical scheme under the present invention and claim scope, it carried out all within protection scope of the present invention.
Claims (6)
1. the searching method for the Video coding estimation is characterized in that comprising following step:
The 1st step: if the absolute difference of initial search point (O) and be less than predetermined threshold, search finishes, otherwise enters diamond search pattern; First calculate rhombus four summits (A), (B), (C) and absolute difference (D) and; If absolute difference and the minimum of initial search point (O), search finishes; If the summit (A) that absolute difference and minimum point are this rhombus, enter the 2nd step;
The 2nd step: relatively (B), (C), (D) and absolute difference (O) with; As (B) or absolute difference (D) be minimum value, enter the 6th step; Otherwise enter the first triangle search pattern, this first triangle search pattern is that the summit (A) of take in the 1st step is unfaithful intention, take (X), (Y) and (Z) be the isosceles triangle on summit; Calculate respectively (X), (Y) and absolute difference (Z) and and with absolute difference and the comparison of (A), as absolute difference and the minimum of (A) point, search end; Absolute difference and minimum as (X) point, enter the 3rd step; Absolute difference and minimum as (Y) point, enter the 4th step; Absolute difference and minimum as (Z) point, enter the 5th step;
The 3rd step: enter the second triangle search pattern, this second triangle search pattern is that to take summit (X) be unfaithful intention, take (B), (Y') and (Z') be the isosceles triangle on summit; Calculate respectively (Y'), (Z') absolute difference and and with absolute difference and the comparison of (X), as absolute difference and the minimum of (X) point, search finishes; As absolute difference and the minimum of (Y') point, return to the 3rd step; Absolute difference and minimum as (Z') point, enter the 4th step;
The 4th step: enter the 3rd triangle search pattern, the 3rd triangle search pattern is that to take summit (Y) be unfaithful intention, take (X'), (Y') and (Z') be the isosceles triangle on summit; Calculate respectively (X'), (Y'), (Z') absolute difference and and with absolute difference and the comparison of (Y), as absolute difference and the minimum of (Y) point, search finishes; As absolute difference and the minimum of (Y') point, return to the 4th step; Absolute difference and minimum as (Z') point, enter the 5th step; Absolute difference and minimum as (X') point, enter the 3rd step;
The 5th step: enter the 4th triangle search pattern, the 4th triangle search pattern is that to take summit (Z) be unfaithful intention, take (D), (X') and (Y) be the isosceles triangle on summit; Calculate respectively (X'), (Y') absolute difference and and with absolute difference and the comparison of (Z), as absolute difference and the minimum of (Z) point, search finishes; As absolute difference and the minimum of (Y') point, return to the 5th step; Absolute difference and minimum as (X') point, enter the 4th step;
The 6th step: enter the 5th triangle search pattern, the 5th triangle search pattern is to take (A), (X) and (Y) be the isosceles triangle on summit; Calculate respectively (X), (Y) absolute difference and and with absolute difference and the comparison of (A), as absolute difference and the minimum of (A) point, search finishes; As absolute difference and the minimum on another summit, enter the 7th step;
The 7th step: enter the 6th triangle search pattern, in the situation that the absolute difference on another summit and minimum, the 6th triangle search pattern is to take this another summit, (A') and (B') be the isosceles triangle on summit; Calculate respectively (A'), (B') absolute difference and and with absolute difference and the comparison on this another summit, as absolute difference and the minimum on this another summit, search finishes; Otherwise repeat the 7th step.
2. the searching method for the Video coding estimation as claimed in claim 1 is characterized in that:
In the 1st step, present encoding piece upper left corner coordinate is added to the coordinate position of motion vectors in reference frame of last encoding block is defined as initial search point (O).
3. the searching method for the Video coding estimation as claimed in claim 2 is characterized in that:
The intermediate value that the motion vectors of encoding block is the adjacent block motion vector, wherein adjacent block comprises three spatial domain adjacent blocks of encoding block described in time domain adjacent block in reference frame and present frame.
4. the searching method for the Video coding estimation as claimed in claim 1 is characterized in that:
In the 1st step, described diamond search pattern is centered by initial search point (O), the step-length rhombus that to be 1 pixel form to left and right, upper and lower expansion.
5. the searching method for the Video coding estimation as claimed in claim 1 is characterized in that:
In the 1st step, described threshold value, for detection of going out static block and accurate static block, is directly ended search to static block and accurate static block.
6. the searching method for the Video coding estimation as claimed in claim 5 is characterized in that:
Described threshold value is 512.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN 201010609255 CN102075748B (en) | 2010-12-27 | 2010-12-27 | Search method for video coding motion estimation |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN 201010609255 CN102075748B (en) | 2010-12-27 | 2010-12-27 | Search method for video coding motion estimation |
Publications (2)
Publication Number | Publication Date |
---|---|
CN102075748A CN102075748A (en) | 2011-05-25 |
CN102075748B true CN102075748B (en) | 2013-12-25 |
Family
ID=44034071
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN 201010609255 Expired - Fee Related CN102075748B (en) | 2010-12-27 | 2010-12-27 | Search method for video coding motion estimation |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN102075748B (en) |
Families Citing this family (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104113756A (en) * | 2013-04-22 | 2014-10-22 | 苏州派瑞雷尔智能科技有限公司 | Integer pixel motion estimation method suitable for H.264 video encoding and decoding |
CN105516725B (en) * | 2015-12-12 | 2018-10-26 | 中国航空工业集团公司西安航空计算技术研究所 | Fine movement estimated data read method based on nine grids |
CN109348234B (en) * | 2018-11-12 | 2021-11-19 | 北京佳讯飞鸿电气股份有限公司 | Efficient sub-pixel motion estimation method and system |
CN112565785B (en) * | 2018-12-28 | 2022-04-26 | 杭州海康威视数字技术股份有限公司 | Coding and decoding method and equipment thereof |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1545325A (en) * | 2003-11-18 | 2004-11-10 | 清华大学 | A coding method based on selfadaptive hexagon search |
CN101009841A (en) * | 2006-01-26 | 2007-08-01 | 深圳艾科创新微电子有限公司 | Estimation method for quick video motion |
CN101031088A (en) * | 2006-02-28 | 2007-09-05 | 展讯通信(上海)有限公司 | Fast cross-movement estimation for realizing video encoder |
EP2046047A1 (en) * | 2007-10-04 | 2009-04-08 | Thomson Licensing | Method and device for performing motion estimation |
-
2010
- 2010-12-27 CN CN 201010609255 patent/CN102075748B/en not_active Expired - Fee Related
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1545325A (en) * | 2003-11-18 | 2004-11-10 | 清华大学 | A coding method based on selfadaptive hexagon search |
CN101009841A (en) * | 2006-01-26 | 2007-08-01 | 深圳艾科创新微电子有限公司 | Estimation method for quick video motion |
CN101031088A (en) * | 2006-02-28 | 2007-09-05 | 展讯通信(上海)有限公司 | Fast cross-movement estimation for realizing video encoder |
EP2046047A1 (en) * | 2007-10-04 | 2009-04-08 | Thomson Licensing | Method and device for performing motion estimation |
Non-Patent Citations (2)
Title |
---|
一种用于视频编码运动估计的新算法;欧阳国胜等;《2006中国西部青年通信学术会议论文集》;20061201;391-394 * |
欧阳国胜等.一种用于视频编码运动估计的新算法.《2006中国西部青年通信学术会议论文集》.2006, |
Also Published As
Publication number | Publication date |
---|---|
CN102075748A (en) | 2011-05-25 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
RU2678510C2 (en) | Method and device for coding and decoding motion vector based on reduced predictors-candidates of motion vector | |
WO2020182165A1 (en) | Method for constructing motion information candidate list and triangular prediction decoding method and apparatus | |
CN101394566B (en) | Cross rhombic motion estimation searching method | |
CN106604035B (en) | A method of the estimation for Video coding and compression | |
CN101420617B (en) | Motion estimation searching method for cross hexagon | |
CN110796662B (en) | Real-time semantic video segmentation method | |
CN102075748B (en) | Search method for video coding motion estimation | |
CN101267556A (en) | Quick motion estimation method and video coding and decoding method | |
CN102801976A (en) | Inter-frame module selecting method based on three-dimensional wavelet video code | |
US9883200B2 (en) | Method of acquiring neighboring disparity vectors for multi-texture and multi-depth video | |
CN102263951B (en) | Quick fractal video compression and decompression method | |
CN103237222B (en) | The method for estimating of multi-mode search | |
CN1791224A (en) | Self-adaptive block searching range rapid motion estimating method based on H.264 | |
CN101621694B (en) | Motion estimation method, motion estimation system and display terminal | |
CN101888546B (en) | A kind of method of estimation and device | |
CN100337482C (en) | Fast motion assessment method based on object edge shape | |
CN107483960B (en) | Motion compensation frame rate up-conversion method based on spatial prediction | |
CN103747262A (en) | Motion estimation method based on GPU (Graphic Processing Unit) | |
CN102447904A (en) | Method for quick motion estimation of video sequences | |
CN101494757A (en) | Motion estimating method based on time-space domain mixing information | |
CN1322758C (en) | Fast motion assessment method based on object texture | |
CN104469380B (en) | Video image forecasting search method based on H.264/AVC standard | |
CN104202606B (en) | One kind determines method based on HEVC estimation starting points | |
CN102263954B (en) | Quick fractal video compression and decompression method based on object | |
CN105306952B (en) | A method of it reducing side information and generates computation complexity |
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 | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20131225 Termination date: 20201227 |
|
CF01 | Termination of patent right due to non-payment of annual fee |