Research On Intra Fast Coding Algorithm For High Efficiency Video Coding

ITU-T and ISO/IEC jointly developed a new generation of video coding standard,called High Efficiency Video Coding(HEVC).Compared with the last generation H.264/AVC,approximately 50% bitrate reduction is achieved for equal perceptual quality.However,the efficient coding performance also brought huge coding complexity,which makes HEVC difficult for rapid and wide adoption.Therefore,reducing coding complexity effectively is a significant research content of HEVC while maintaining almost the same video quality.The improved algorithms are proposed to reduce coding complexity from prediction mode decision and Coding Unit(CU)partitioning in intra coding.In order to address the high computation complexity problem of prediction mode decision in intra coding,a structure similarity and gradient intensity based fast prediction mode decision algorithm(SS-GI-MD)is proposed for HEVC intra coding.In the SS-GI-MD algorithm,for skipping the rough mode decision(RMD)and reducing candidate mode numbers of the rate distortion optimization(RDO),the relations of Prediction Unit(PU)are fully exploited by the structure similarity of nearby 4×4PUs.As a result,the prediction mode decision process is improved.Furthermore,the intensity gradient is utilized to select numbers of prediction mode to be involved in the RMD so that the encoding time can be reduced.The experimental results show that the SS-GI-MD algorithm achieves a 24% time saving of intra encoding on average while maintaining almost the same video quality as the HM14.0.Considering the problem of high computation complexity in partitioning process of CU,a texture complexity and depth prediction based fast CU decision algorithm(FCU-TC-DP)is proposed for HEVC intra coding.The texture complexity is used in advance to determine that Largest Coding Unit(LCU)is split or not.Then,the depth level of current CU is predicted by fully utilizing depths of coded neighboring CUs to reduce the computing on redundant depth level.Compared with the HM14.0,experimental result shows that the FCU-TC-DP algorithm can save a 38% intra encoding time on average with the quality of video basically unchanged.In conclusion,the improved algorithms of intra prediction mode decision and CU partitioning achieve competitive results while maintaining almost the same quality of video.With negligible loss of coding efficiency,the coding complexity is effectively reduced.