Research On Coding Algorithm For H.265/HEVC

In recent years,high-definition(HD)and ultra-high-definition(UHD)video applications are widely used in various areas.The video coding standards proposed previously have been unable to satisfy the demand of encoding.In this case,the newest video coding standard named High Efficiency Video Coding(HEVC)came into being.The new technologies in the coding framework of HEVC make the coding efficiency greatly improved,but they also increase the complexity of coding and hinder the real-time application of HEVC.Therefore,it is very important to reduce the coding complexity without affecting the compression efficiency.Firstly,we introduce the basic principle of video encoding,the development history of the video encoding standard,the key technologies and the cost function of HEVC.Then,we analyze the processes of block partitioning and SAO(Sample Adaptive Offset)mode selection of HEVC.At last,a fast algorithm of partitioning algorithm and a fast SAO mode selection algorithm are proposed.The main contributions of this dissertation are as follows:(1)A fast algorithm for block partitioning based on Bayesian decision is proposed.Firstly,the scene change detection based on average grey difference is used to divide the video sequence into the online learning phase and the fast division phase.Secondly,for the online learning phase and the fame in which scene changes,the Jinter and Jintra are extracted from CUs in every depth to establish the Gauss mixture models.Then the EM(Expectation Maximazation)algorithm initialized by K-Means algorithm is utilized to determine the specific parameters.For CUs in the fast division stage,the characteristic values are extracted,and then the conditional probabilities for CUs to the decisions of “division” and “no division” are calculated.Finally,the minimum risk Bayesian decision rule is used to figure out the decision with smaller risk,and the decision is regarded as the judement of the current CU.The numerical results based on the proposed algorithm tested on various video sequences show a reduction of the encoding time by about 44.5% over the original HM13.0 algorithm,while the encoding bit rate BD-Rate(Bj?ntegaard Delta Bit Rate)increases only 0.94%.(2)A fast algorithm for SAO mode selection based on the dominant edge(DE)is proposed.Firstly,for each CTU(Coding Tree Unit),the DE is detected by the dominant edge assent(DEA).Secondly,the corresponding edge offset(EO)mode is regarded as the optimal EO for the current CTU.Finally,the Rate Distortion costs(RD Costs)of the optimal EO mode,band offset(BO)mode and without using SAO are compared,and the mode with minimum RD cost is selected as the optimal SAO mode.The fast algorithm saves 53.4% of the SAO time over the original HM13.0 algorithm,and the coding performance loss is small.