CN101795409B - content adaptive fractional pixel motion estimation method - Google Patents
content adaptive fractional pixel motion estimation method Download PDFInfo
- Publication number
- CN101795409B CN101795409B CN 201010117539 CN201010117539A CN101795409B CN 101795409 B CN101795409 B CN 101795409B CN 201010117539 CN201010117539 CN 201010117539 CN 201010117539 A CN201010117539 A CN 201010117539A CN 101795409 B CN101795409 B CN 101795409B
- Authority
- CN
- China
- Prior art keywords
- search
- pixel motion
- vector
- pred
- point
- 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
Landscapes
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
The invention belongs to the field of video coding in signal processing, and relates to a content adaptive fractional pixel motion estimation method, which adopts a method for omitting the invalid fractional pixel motion vector search based on flat-site macro block prediction and is characterized in that: whether the current macro block is a flat block or not is judged by detecting whether the motion vector of the mode 1 of seven H.264 motion estimation modes falls on the integer pixel point or not. The flat block is only subjected to routine integer pixel motion vector search by the follow-up mode; and the non-flat block is subjected to routine integer pixel and fractional pixel motion vector search. The improved enhancement diamond template searching method based on the predicted vector adopts an improved searching template to make the search stop in the range of [-2, 2] of the predicted motion vector, and omits to calculate few pairs of fractional pixel sampling points which are out of the range of [-2, 2] and have low code-efficiency increase. Compared with the full fractional pixel search method (FFPS) and under the condition that the PSNR (peak signal to noise ratio) has tiny reduction (0.095 to 0.209dB), the content adaptive fractional pixel motion estimation method averagely reduces 75.6 percent of fractional pixel search points, and averagely saves 38.5 percent of calculated quantity of the whole motion estimation module. The content adaptive fractional pixel motion estimation method of the invention can reduce the calculated quantity of the fractional pixel motion estimation on the basis of ensuring the accuracy of motion estimation.
Description
Technical field
The invention belongs to the field of video encoding in the signal processing, H.264/AVC proposed new content adaptive fractional pixel motion estimation method at up-to-date international video encoding standard especially.Can on the basis of guaranteeing the estimation precision, reduce the amount of calculation of fraction pixel estimation.
Background technology
H.264/AVC be the up-to-date video coding international standard of formulating by ITU-T and the common joint video team JVT (Joint Video Team) that sets up of ISO/IEC.In video coding system H.264, estimation can effectively be removed the time redundancy of video sequence consecutive frame, has determined coding rate, compression ratio and the decoded video quality of video encoder to a great extent.Therefore, H.264 motion estimation module has increased multiple coding techniques, for example, and 1/4 pixel prediction precision, multi-reference frame, the motion compensation of tree.H.264/AVC performance has surmounted all video encoders in the past, under the prerequisite of same-code quality, H.264/AVC the code check that produces of Baseline Profile more about 40% than H.263Baseline having saved (referring to
Ostermann, Jan Bormans, Peter List, Detlev Marpe, Matthias Narroschke, Fernando Pereira, Thomas Stockhammer, Thomas Wedi.Video coding with is H.264/AVC:tools, performance, and complexity[J] .IEEE Circuits and Systems Magazine, 2004,4 (1): 7-28.).H.264 high-performance is the cost that rises to computation complexity, its computation complexity approximately is H.263 4 to 5 times, wherein, the amount of calculation of motion estimation module accounts for the 50%-90% of whole encoder (referring to Yu-Wen Huang, Ching-Yeh Chen, Chen-Han Tsai, Chun-Fu Shen, Liang-Gee Chen.Survey on block matching motion estimation algorithms and architectures with new results[J] .Journal of VLSI Signal Processing Systems for Signal, Image, and Video Technology, 2006,42 (3): 297-320.).
H.264 in the encoder, estimation comprises two parts: whole pixel motion is estimated and the fraction pixel estimation.The fraction pixel estimation is in the basic enterprising row interpolation computing that obtains whole pixel motion vector (MV:motion vector), and search obtains the motion vector process of fraction pixel precision.The fraction pixel estimation can improve the performance of encoder greatly on compressed video quality and compression ratio.Experimental result shows, uses fractional pixel motion estimation method than only using whole pixel motion estimating method compression ratio and on average improves 48%, and simultaneously, Y-PSNR (PSNR) improves 1~3dB.But, because extra computing, as, interpolation and fraction pixel search, the fraction pixel estimation has increased the amount of calculation of whole motion estimation module greatly.
Whole pixel motion estimating method is a hot research in recent years.H.264 the JM reference software of standard has adopted two kinds of fast whole pixel motion estimating method UMHexagonS (referring to Zhibo Chen, Peng Zhou, Yun He, Yidong Chen.Fast integer pel and fractional pel motion estimation for JVT[C], JVT-F017,2002.) and EPZS (referring to Alexis Michael Tourapis, Hye-Yeon Cheong, Pankaj Topiwala.Fast ME in the JM reference software[C], JVT-P026,2005.), compare with all direction search method, greatly reduce search point amount (average search of each motion vector is counted and reduced to below 10), computation complexity reduces more than 90%.H.264 adopt the tree motion compensation, have the estimation of 7 kinds of patterns, if adopt traditional 1/4 precision fraction pixel all direction search method (full fractional pixel search), each pattern needs 16 search points, and then each macro block need be searched for 112 points.Therefore, the improvement of fraction pixel precision method for estimating becomes the key that whole motion estimation module is optimized.
H.264/AVC two kinds of fractional pixel motion estimation methods have been adopted at present, fraction pixel all direction search method (FFPS:Full Fractional Pixel Search) and based on the fraction pixel searching method (CBFPS:Center Based Fractional Pixel Search) at center.(referring to Zhibo Chen, Peng Zhou, Yun He, Yidong Chen.Fast integer pel and fractional pel motion estimation for JVT[C], JVT-F017,2002.)
The FFPS method as shown in Figure 1.FFPS is that the level search is carried out at the center with the whole pixel of the best: at first, 8 1/2 location of pixels around the whole location of pixels of calculating optimum find best 1/2 pixel match point; Then, 8 1/4 location of pixels around calculating optimum 1/2 pixel find best 1/4 pixel match point, as the optimum movement vector of FFPS.FFPS need calculate 16 location of pixels.
The CBFPS method as shown in Figure 2.CBFPS adopts the vector prediction and the search strategy of refinement progressively: at first, the intermediate value of the motion vector by calculating adjacent block obtains fraction pixel motion vectors (Pred_x, Pred_y), more original search center (0,0) and motion vectors (Pred_x, Pred_y) matching error, the search point that wherein produces minimal error is as minute starting point of pixel search; Then, use the rhombus template (referring to Jo Yew Tham, Surendra Ranganath, Maitreya Ranganath, Ashraf Ali Kassim.A novel unrestricted center-biased diamond search algorithm for block motion estimation[J] .IEEE Transactions on Circuits and Systems for Video Technology, 1998,8 (4): 369-377.) iteration, with progressively refinement, calculate final fraction pixel match point.Compare with FFPS, CBFPS can reduce by 20% amount of calculation, and picture quality is almost constant, is a kind of very effective rapid fraction pixel searching method.
H.264/AVC the inter prediction of standard provided greater flexibility than former coding standard.Adopt the motion compensation of tree, each macro block can be divided into 16 * 16,8 * 16,16 * 8,8 * 8 piece; When adopting 8 * 8, can also be further divided into 8 * 4,4 * 8,4 * 4 littler sub-pieces, as shown in Figure 3.
When carrying out estimation, encoder need { SKIP (adopts the SKIP pattern, with the consistent zone with motion of the large-area stagnant zone of encoding effectively to whole predictive mode set.The SKIP pattern also is the motion compensated prediction pattern of 16 * 16 sizes, and any motion and prediction residual information just do not need to encode), 16 * 16,8 * 16,16 * 8,8 * 8,8 * 4,4 * 8,4 * 4} is cut apart at every kind and is carried out estimation under the pattern separately, obtains motion vector separately.Encoder is taken all factors into consideration encode required bit number and to the motion vector required bit number of encoding of residual values, selects best predictive mode then.H.264/AVC select the method for optimization model be called rate-distortion optimization (Rate Distortion Optimization, RDO).By the coded macroblocks under each candidate pattern and the rate distortion costs of gained relatively, the pattern of choosing the rate distortion costs minimum is as the forced coding pattern based on the mode selecting method of RDO.
The rate distortion costs function definition is as follows:
J(s,c,MODE|QP,λ
MODE)=SSD(s,c,MODE|QP)+λ
MODE·R(s,c,MODE|QP)(1)
Wherein, Qp is quantization parameter (Quantization Parameter), and MODE is candidate's macro-block coding pattern, to I frame and P frame, Lagrangian factor lambda
MODE=0.85 * 2
QP/3SSD is the mean square deviation sum of primary signal s and reconstruction signal c, as distortion metrics.R is quantization parameter QP and the pattern MODE required bit number of this macro block of coding (comprise behind macro block header, motion vector and the residual error change quantization coefficient etc.) down.
Figure 4 shows that the rate distortion costs computational process under a certain candidate pattern, as seen, rate distortion costs for calculated candidate pattern correspondence, need motion estimation/compensation, conversion, quantification and entropy coding process to obtain the required bit number of coding under this pattern, also need inverse quantization/inverse transformation to obtain reconstruction signal, whole process has very high computation complexity.By formula (1) as seen, rate distortion costs is the compromise of distortion and required bit number, and the shared weight of bit number is the Lagrangian factor, and its value is the monotonic function of quantization parameter.
Summary of the invention
The objective of the invention is to overcome the weak point of prior art, propose a kind of content adaptive fractional pixel motion estimation method.Can on the basis of guaranteeing the estimation precision, reduce the amount of calculation of fraction pixel estimation.This method is applicable to H.264 international standard, but is not limited to H.264, can extend to the application of other video compression international standard and non-international standard.
The present invention proposes a kind of content adaptive fractional pixel motion estimation method.It is characterized in that, adopt invalid fraction pixel motion-vector search omission method and improved enhancement mode rhombus template searching method based on predictive vector based on the flat site macroblock prediction.
Invalid fraction pixel motion-vector search omission method based on the flat site macroblock prediction is specific as follows:
Whether drop on whole pixel by the motion vector that detects the pattern 1 in 7 kinds of patterns of estimation H.264, judge whether current macro is flat block.For flat block, follow-up mode is only carried out the whole pixel motion vector search of routine, and does not carry out the fraction pixel motion-vector search.For the non-flat forms piece, carry out whole pixel of routine and fraction pixel motion-vector search, at fraction pixel motion-vector search wherein, adopt following improved enhancement mode rhombus template searching method based on predictive vector.
Improved enhancement mode rhombus template searching method based on predictive vector is specific as follows:
The first step: the fraction pixel motion vector by adjacent block prediction current block, obtain FMVP, promptly (Pred_x, Pred_y).Directly with FMVP as the search starting point.
Second step: comparison search starting point (Pred_x, Pred_y) 4 diamond search points on every side and (Pred_x, matching error Pred_y), if least absolute error sum MSAD is positioned at (Pred_x, Pred_y), then stop the fraction pixel motion-vector search, otherwise carry out the search of the 3rd step.
The 3rd step: if optimal match point is relative with the suboptimum match point, then selecting optimal match point MV is final fraction pixel motion vector; If optimal match point is adjacent with the suboptimum match point, then calculate the matching error of putting on the square template that is adjacent, if MSAD still is the rhombus optimal match point, then selecting rhombus optimal match point MV is final fraction pixel motion vector, otherwise carries out next step.
The 4th step: with the search point on the square template in the 3rd step is the center, searches for its point on every side with the rhombus template.The point of selecting MASD is as final fraction pixel motion vector.
The advantage that the present invention is compared with prior art had is: the invalid fraction pixel motion-vector search omission method based on the flat site macroblock prediction of the present invention has made full use of the motion estimation mode correlation.According to the motion vector of pattern 1, prediction flat block SMB.For SMB, all the other 6 kinds of match patterns are only put in order the pixel search, skip the search of branch pixel.Experiment shows that this method under the prerequisite that guarantees decoded image quality, can reduce 28.70%~56.00% with the amount of calculation of fraction pixel motion-vector search.This method has independence, and the enhancement mode rhombus template searching method based on predictive vector that proposes with this paper is used in combination, and can further reduce the amount of calculation of fraction pixel estimation.Content adaptive fractional pixel motion estimation method is compared with best practice FFPS, on average saves for 38.5% computing time, and the PSNR loss is no more than 0.209dB.Especially,, can save for 46% computing time, and keep suitable with it even more excellent PSNR for the mild video sequence of motion.
Description of drawings
Fig. 1 searches for (FFPS) schematic diagram entirely for fraction pixel.
Fig. 2 is rapid fraction pixel search (CBFPS) schematic diagram based on the center.
Fig. 3 is variable macroblock size schematic diagram H.264.
Fig. 4 is the computational process of the rate distortion costs under a certain candidate pattern.
Fig. 5 is 7 kinds of spatial relationships between the pattern.
Fig. 6 is search pattern commonly used: (a) be the rhombus template; (b) be square template; (c) be the hexagon template.
Fig. 7 is the enhancement mode rhombus template search schematic diagram based on predictive vector.
Fig. 8 is the invalid fraction pixel motion-vector search omission method flow diagram based on the flat site macroblock prediction.
Embodiment
The content adaptive fractional pixel motion estimation method that the present invention proposes reaches embodiment in conjunction with the accompanying drawings and is described in detail as follows:
The content adaptive fractional pixel motion estimation method that the present invention proposes comprises: based on the invalid fraction pixel motion-vector search omission method and the improved enhancement mode rhombus template searching method based on predictive vector of flat site macroblock prediction.Introduce respectively below:
Invalid fraction pixel motion-vector search omission method based on the flat site macroblock prediction may further comprise the steps:
The fraction pixel estimation can improve compressed video quality and compression ratio, and still, the fraction pixel motion-vector search needs huge amount of calculation.If the MSAD that the least absolute error that the fraction movement Vector search obtains and (MSAD:minimum sum of absolute difference) obtain greater than whole pixel motion vector search, then select whole pixel MV as final MV, it is invalid that the fraction pixel motion-vector search is regarded as.Otherwise the fraction pixel motion-vector search is effectively, selects fraction pixel MV as final MV.Preceding 20 frames to 7 standard video sequence experimentize, and (wherein, News, Container, Silent belong to the low spatial details and the cycle tests slowly that moves; Paris, Foreman are medium spatial detail and the medium cycle tests of exercise intensity; Football is the high spatial details and the violent cycle tests that moves), be that the whole shared ratio of pixel MV is listed in table 1 with final MV.
The shared ratio of the whole pixel MV of table 1 (10 frames, QP=28)
The sequence that the video content motion is mild as News, Container, Silent, has the motion vector above 80% to be positioned at whole location of pixels.In fact, for the flat site of frame of video, the fraction pixel search is also not obvious to the raising of coding efficiency, is no efficient search.If can predict these flat area block, skip invalid fraction pixel search, just can reduce unnecessary calculating.
Therefore, how to make encoder whether carry out the fraction pixel search according to video content adaptive ground decision, that is, how before the fraction pixel estimation, the flat area block of predicted video frame is a key point of omitting invalid fraction pixel motion estimation vectors search.
H.264 there is stronger correlation (7 kinds pattern between spatial relationship such as Fig. 5) between 7 kinds of motion estimation modes.Utilize the result of upper mode motion-vector search can predict the smooth degree of current macro.Therefore, if upper mode 1 (16 * 16) motion vector at whole location of pixels, then defining such macro block is flat block, is called for short SMB (smooth macro-block).For SMB, the motion-vector search of following layer model can be skipped the fraction pixel motion-vector search.
The invalid fraction pixel motion-vector search omission method based on the flat site macroblock prediction that the present invention proposes is called for short SMBP (smooth macro-block prediction).Whether drop on whole pixel by detecting pattern 1 motion vector, judge whether current macro is flat block.For flat block, follow-up mode is only carried out the whole pixel motion vector search of routine, and does not carry out the fraction pixel motion-vector search.For the non-flat forms piece, carry out whole pixel of routine and fraction pixel motion-vector search.Flow chart such as Fig. 8.
The flat block predictablity rate refers to that the following layer model that is judged as SMB carries out conventional fraction pixel motion-vector search, and its final motion vector drops on the ratio of whole location of pixels.Accuracy rate such as formula (2) definition.Flat block predictablity rate, whole pixel MV proportion statistics see Table 2.
Predictablity rate is high more, and matching error is more little, so decoded image quality decline is more little, and code check changes more little.By table 2 as seen, for the video sequence slowly that moves, the accuracy rate of flat block prediction is more than 91%, and the inventive method can be made prediction more accurately to this type of sequence.The ratio that fraction pixel MV reduces is from 28.70%~56.00%, that is, the inventive method can reduce 28.70%~56.00% with the amount of calculation of fraction pixel motion-vector search.
Table 2 flat block predictablity rate (10 frames, QP=28)
Improved enhancement mode rhombus template searching method based on predictive vector may further comprise the steps:
Fraction pixel is obtained by whole picture element interpolation, and the correlation of search point is far above the correlation of whole pixel search positions in the fraction pixel search window.When searching for point near overall smallest point, matching error is dull to descend.Therefore, many rapid fraction pixel motion vector searching methods have adopted motion vectors (FMVP:fractional predicted mv) as the search starting point.If initial point that can accurately predicting fraction pixel motion-vector search then can earlier search near the best MV the FMVP, in time stop the fraction pixel motion estimation search.
The FMVP of current block is by the intermediate value decision of the fraction pixel motion vector of adjacent block (upward, the left and right piece of going up).FMVP comprises two parts information: whole pixel prediction vector and fraction pixel predictive vector.Extract the fraction pixel predictive vector with formula (3), wherein mv has searched for the whole pixel motion vector that obtains, is unit with the fraction pixel.% is a modulo operation, β=4 o'clock, and search precision is 1/4 pixel, β=8 o'clock, search precision is 1/8 pixel.
frac_pred_mv=(pred_mv-mv)%β(3)
Table 3 has shown FMVP and the matching degree of the best score pixel MV that obtained by FFPS.Coupling expression FMVP equals best MV; Distance between [1,1] expression FMVP and the best MV is within 1 fraction pixel unit.
Can find to move the matching probability of the FMVP of cycle tests slowly and best MV greater than 82% by table 3.In the matching probability of cycle tests of high exercise intensity lower, but the probability of its FMVP in best MV [2,2] scope is more than 90%.Therefore, can use the starting point of FMVP as the fraction pixel motion-vector search.
The matching degree of table 3 FMVP and best score pixel MV
Motion-vector search is used three kinds of templates always: rhombus template, square template and hexagon template.Wherein, the rhombus template is the simplest, is adopted by many video encoders, as Fig. 6 (a); The square template has increased by 4 points on the diagonal on the rhombus template, computation complexity and Search Results accuracy increase, as Fig. 6 (b); Hexagon is fit to the bigger occasion in hunting zone, because the fraction pixel motion vector search ranges only limits between two whole pixels, the hexagon template is not suitable for the fraction pixel motion-vector search, as Fig. 6 (c).
The present invention proposes the enhancement mode rhombus template searching method based on predictive vector.Different with CBFPS is, because FMVP and best MV have higher matching rate, this method is not considered original search center (0,0), and directly with FMVP as the search starting point; Adopt enhancement mode rhombus template (EDSP:extended diamond search pattern), in conjunction with square template accuracy advantage of higher, at the search point that increases on the basis of rhombus template on the diagonal; Do not carry out the iteration of rhombus template, and search is stopped in [2,2] scope of FMVP, omit [2,2] extraneous minority code efficiency is improved little fraction pixel motion-vector search,, thereby further reduce amount of calculation with the minimizing search point.
Fig. 7 is the enhancement mode rhombus template search strategy schematic diagram based on predictive vector, and method flow is as follows.
The first step: the fraction pixel motion vector by adjacent block prediction current block, obtain FMVP, promptly (Pred_x, Pred_y).Directly with FMVP as the search starting point.
Second step: comparison search starting point (Pred_x, Pred_y) 4 diamond search points on every side and (Pred_x, matching error Pred_y), if least absolute error sum MSAD is positioned at (Pred_x, Pred_y), then stop the fraction pixel motion-vector search, otherwise carry out the search of the 3rd step.
The 3rd step: as Fig. 7 (a), if optimal match point is relative with the suboptimum match point, then selecting optimal match point MV is final fraction pixel motion vector; As Fig. 7 (b), if optimal match point is adjacent with the suboptimum match point, then calculate the matching error of putting on the square template that is adjacent, if MSAD still is the rhombus optimal match point, then selecting rhombus optimal match point MV is final fraction pixel motion vector, otherwise carries out next step.
The 4th step: with the search point on the square template in the 3rd step is the center, searches for its point on every side with the rhombus template.The point of selecting MASD is as final fraction pixel motion vector.
H.264, the content adaptive fractional pixel motion estimation method that the present invention proposes is being tested on the test platform JM15.0.Selected representative motion severe degree from slowly testing to 6 violent video sequences.The parameter setting relevant with estimation of JM15.0 encoder sees Table 4.
Table 4 encoder relative parameters setting
With EDSP, SMBP+EDSP contrasts on the method performance with FFPS and the CBFPS that JM15.0 adopts respectively: (1) search point (fraction pixel): each macro block obtains the search point that final fraction pixel motion vector needs, the matching speed of reflection method; (2) variation of Y-PSNR (Δ PSNR:Peak Signal Noise Radio): the image after weighing estimation and compensating and the difference of original image, the forecast quality of reflection method; (3) Δ computing time (comprise whole pixel and fraction pixel the computing time that estimation is saved): reflect the time that encoder consumes on the whole on motion estimation module.These three experiment parameters all are that standard compares with FFPS.
Each method performance of table 5 relatively
By table 5 as seen, average each the macro block search point of the EDSP method that this paper proposes is 3.9, has saved 75.6% than FFPS; The PSNR of EDSP method compares with FFPS, and loss is no more than 0.13dB.SMBP can further reduce operand in conjunction with EDSP, saves operation time.SMBP+EDSP compares with EDSP, for the mild video sequence of motion, as News, Container, Silent, on average saves for 16% computing time; The video sequence that exercise intensity is medium as Paris and Forman, on average saves for 10% computing time; The violent video sequence that moves as Football, saves for 7% computing time.SMBP+EDSP compares with best practice FFPS, on average saves for 38.5% computing time, and the PSNR loss is no more than 0.209dB; SMBP+EDSP compares with CBFPS, on average saves for 14.7% computing time, and the PSNR loss is no more than 0.196dB.
Claims (1)
1. invalid fraction pixel motion-vector search omission method that is used for the employing of content adaptive fractional pixel motion estimation method based on the flat site macroblock prediction, whether drop on whole pixel by the motion vector that detects the pattern 1 in 7 kinds of patterns of estimation H.264, judge whether current macro is flat block; For flat block, follow-up mode is only carried out the whole pixel motion vector search of routine, and does not carry out the fraction pixel motion-vector search; For the non-flat forms piece, carry out whole pixel of routine and fraction pixel motion-vector search, it is characterized in that:
At fraction pixel motion-vector search wherein, adopt improved enhancement mode rhombus template searching method based on predictive vector, carry out following four steps:
The first step: the fraction pixel motion vector by adjacent block prediction current block, obtain motion vectors FMVP, promptly (Pred_x, Pred_y); Directly with FMVP as the search starting point;
Second step: comparison search starting point (Pred_x, Pred_y) point of the search of 4 on the rhombus template on every side and (Pred_x, Pred_y) matching error, if least absolute error sum MSAD is positioned at (Pred_x, Pred_y), then stop invalid fraction pixel motion-vector search omission method, otherwise carry out the search of the 3rd step;
The 3rd step: if optimal match point is relative with the suboptimum match point, then selecting optimal match point MV is final fraction pixel motion vector; If optimal match point is adjacent with the suboptimum match point, then calculate the matching error of putting on the square template adjacent with optimal match point, if MSAD still is the rhombus optimal match point, then selecting rhombus optimal match point MV is final fraction pixel motion vector, otherwise carries out next step;
The 4th step: with the search point on the square template in the 3rd step is the center, searches for its point on every side with the rhombus template; The point of selecting MSAD is as final fraction pixel motion vector.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN 201010117539 CN101795409B (en) | 2010-03-03 | 2010-03-03 | content adaptive fractional pixel motion estimation method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN 201010117539 CN101795409B (en) | 2010-03-03 | 2010-03-03 | content adaptive fractional pixel motion estimation method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN101795409A CN101795409A (en) | 2010-08-04 |
CN101795409B true CN101795409B (en) | 2011-12-28 |
Family
ID=42587797
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN 201010117539 Expired - Fee Related CN101795409B (en) | 2010-03-03 | 2010-03-03 | content adaptive fractional pixel motion estimation method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN101795409B (en) |
Families Citing this family (15)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102625293B (en) | 2012-03-21 | 2014-11-05 | 华为技术有限公司 | Method, device and system for notifying and acquiring address information invalidation |
US9325991B2 (en) * | 2012-04-11 | 2016-04-26 | Qualcomm Incorporated | Motion vector rounding |
CN102740073B (en) | 2012-05-30 | 2015-06-17 | 华为技术有限公司 | Coding method and device |
CN103281533B (en) * | 2013-05-14 | 2016-02-24 | 芯原微电子(北京)有限公司 | For equipment and the method for enhancement layer estimation in scalable video |
CN103873875A (en) * | 2014-03-25 | 2014-06-18 | 山东大学 | Layering sub pixel motion estimation method for image super resolution |
CN103957420B (en) * | 2014-04-30 | 2017-02-15 | 华南理工大学 | Comprehensive movement estimation modified algorithm of H.264 movement estimation code |
CN104811728B (en) * | 2015-04-23 | 2018-03-02 | 湖南大目信息科技有限公司 | A kind of method for searching motion of video content adaptive |
CN105611299B (en) * | 2015-12-25 | 2018-11-23 | 北京工业大学 | A kind of method for estimating based on HEVC |
CN105872310B (en) * | 2016-04-20 | 2020-03-17 | 上海联影医疗科技有限公司 | Image motion detection method and image noise reduction method for movable imaging equipment |
WO2020006690A1 (en) * | 2018-07-03 | 2020-01-09 | 深圳市大疆创新科技有限公司 | Video processing method and device |
CN109089121B (en) * | 2018-10-19 | 2021-06-22 | 北京金山云网络技术有限公司 | Motion estimation method and device based on video coding and electronic equipment |
CN109660811B (en) * | 2018-12-17 | 2020-09-18 | 杭州当虹科技股份有限公司 | Rapid HEVC inter-frame coding method |
JP7269371B2 (en) * | 2019-04-25 | 2023-05-08 | ベイジン・ダジア・インターネット・インフォメーション・テクノロジー・カンパニー,リミテッド | Method and Apparatus for Prediction Improvement Using Optical Flow |
JP7381615B2 (en) | 2019-06-25 | 2023-11-15 | オッポ広東移動通信有限公司 | Motion compensation processing method, encoder, decoder, storage medium |
CN112911310B (en) * | 2021-01-15 | 2023-05-16 | 北京博雅慧视智能技术研究院有限公司 | Multi-layer whole pixel motion estimation searching method, device, equipment and storage medium |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101394566A (en) * | 2008-10-29 | 2009-03-25 | 北京航空航天大学 | Cross rhombic motion estimation searching method |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7580456B2 (en) * | 2005-03-01 | 2009-08-25 | Microsoft Corporation | Prediction-based directional fractional pixel motion estimation for video coding |
CA2664668C (en) * | 2006-10-10 | 2014-04-01 | Nippon Telegraph And Telephone Corporation | Intra prediction encoding control method and apparatus, program therefor, and storage medium which stores the program |
-
2010
- 2010-03-03 CN CN 201010117539 patent/CN101795409B/en not_active Expired - Fee Related
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101394566A (en) * | 2008-10-29 | 2009-03-25 | 北京航空航天大学 | Cross rhombic motion estimation searching method |
Also Published As
Publication number | Publication date |
---|---|
CN101795409A (en) | 2010-08-04 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN101795409B (en) | content adaptive fractional pixel motion estimation method | |
CN101448159B (en) | Rapid interframe mode selection method based on rate-distortion cost and mode frequency | |
CN101640802B (en) | Video inter-frame compression coding method based on macroblock features and statistical properties | |
CN100551075C (en) | A kind of method for choosing frame inner forecast mode of low complex degree | |
JP5797634B2 (en) | High speed geometric mode determination method and apparatus in video encoder | |
CN101023674B (en) | Method and apparatus for fast mode decision for interframes | |
CN102740077B (en) | H.264/AVC standard-based intra-frame prediction mode selection method | |
CN100468982C (en) | Method and apparatus for performing high quality fast predictive motion search | |
CN103188496B (en) | Based on the method for coding quick movement estimation video of motion vector distribution prediction | |
CN103248895B (en) | A kind of quick mode method of estimation for HEVC intraframe coding | |
CN103384325A (en) | Quick inter-frame prediction mode selection method for AVS-M video coding | |
CN103873861A (en) | Coding mode selection method for HEVC (high efficiency video coding) | |
CN103384327A (en) | AVS fast mode selection algorithm based on adaptive threshold | |
CN102158709A (en) | Derivable motion compensation prediction method of decoding terminal | |
CN101394566A (en) | Cross rhombic motion estimation searching method | |
CN103546758A (en) | Rapid depth map sequence interframe mode selection fractal coding method | |
CN103384326A (en) | Quick intra-frame prediction mode selection method for AVS-M video coding | |
CN104168480A (en) | Fast selecting method for intra-frame prediction encoding modes based on HEVC standard | |
CN102291582A (en) | Distributed video encoding method based on motion compensation refinement | |
CN101754022A (en) | Motion estimation method with low complexity | |
CN105187826A (en) | Rapid intra-frame mode decision method specific to high efficiency video coding standard | |
CN101304529A (en) | Method and device for selecting macro block pattern | |
CN102164283A (en) | Sub-pixel motion estimation method based on AVS (audio video standard) | |
CN101883275B (en) | Video coding method | |
JP5566786B2 (en) | Error absolute value sum estimation system and estimation 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 | ||
C17 | Cessation of patent right | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20111228 Termination date: 20140303 |