Motion Estimation Algorithm Research Of Inter Prediction In High Efficiency Video Coding

With the constant revolution of information and multimedia technology innovation,H.264/AVC video coding standard is no longer able to meet the demand of currentsociety. A new generation of Video Coding Standard HEVC is jointly formulated OnJanuary26,2013by ITU-T VCEG (Video Coding Experts Group) and ISO/IEC(International Standard Organization/International Electro Technical Commission). Itdoes better on compression performance, but it needs more decoding time. The furtherresearch of the motion estimation search strategy and matching principles consumingmost time which is the core of the motion estimation two key factors in video codingtechnology is necessary.This thesis introduces the theory, the new technology and the new concept ofHEVC in detail firstly, verify the improvement of the compression performance and thepreponderance ultra to high-definition video when comparing to the H.264/AVCthrough the experiment. And then introduced the related principle of inter prediction ofHEVC, and proposes a difference discriminated classification method based on theprediction process, which short the encoding time by more than0.7%. Finally the searchstrategy and matching principles which are the two aspects of the block matchingmotion estimation algorithm is studied. In terms of search strategy, this paper analyzesthe basic theory of the general motion estimation search strategy, and throughexperiment it compared the HEVC standard hybrid search strategy; In terms ofmatching criterion, based on the study of the general matching criterion, this paperproposes a sparse statistical rule of secondary Renyi entropy matching motionestimation, when calculating the secondary Renyi entropy, it introduces statisticalhistogram, which using statistical methods to calculate the probability density function,and combined with image quality assessment based on gradient and motion vectorcenter deviation characteristics, in order to sparse the statistical interval of the histogram.Through experiment simulation, this article demonstrates that the proposed matchingcriterion for dramatic video sequence can do better than the traditional absolute errorfunction on the value of PSNR.