Study On Fast Intra Prediction Algorithm For High Efficiency Video Coding

High Efficiency Video Coding HEVC is the latest video compression standard which can save bit-saving by 50% compared to its predecessors of H.264/AVC High profile with similar visual quality.In HEVC,35 modes are proposed for intra prediction to improve the compression efficiency.But,at the same time,it caused high computational to encoder.Hence,How to reduce the complexity of the HEVC coding reasonably has become a research hotspots in the field of video coding.In this paper,we propose two fast intra prediction algorithms to the high computational part of HEVC,Which is “The Fast Intra Prediction Block Partitioning Algorithm Based On Image Correlation” and “The Fast Intra Prediction Algorithm Based On Reduced Number Of Prediction Modes”.Firstly,a most possible depth interval is proposed according to the correlation analysis of adjacent CTU depth range and the adjacent frame.And in this interval,CU is divided into small size.This method could save encoding time by early terminate CU partition or skip some unnecessary division of CU.Experimental results demonstrate that,under the prerequisite of video quality guaranteed,the fast intra prediction block partitioning algorithm based on image correlation can reduce 27.5% intra coding time with 0.34% BD-rate raise and 0.03 d B PSNR loss on average.And then,we reduce the number of prediction modes in RMD from 35 to 17 according to the directional of angle mode.Meanwhile,we further reduce the number of candidate modes in subsequent RDO process and reduce complexity.Experimental results show that,under the prerequisite of video quality guaranteed,the fast intra prediction algorithm based on reduced number of prediction modes can reduce 26.89% intra coding time with 0.06%BD-rate raise and 0.02 d B PSNR loss on average.