Research On High Efficiency Prediction In Video Compression

Video has been known by everybody and became an indispensable part of people’s life.With the increasing of people’s sensory requirements,high-definition video and three dimensions video are gradually becoming mainstream.While,high quality of video also requires more storage space and transmission bandwidth.In recent years,the common compression standards are MPEG-4,H.264,HEVC,and so on,using the image correlation to remove the redundant information.AVS2 is the second generation of Audio and Video Coding Standard searched by China independently,and its coding efficiency almost equals to the international coding standard HEVC.Meanwhile,3D-AVS for multi-view video completed in 2014,for further study of 3D video compression.This paper researched the tradition compression standard HEVC and 3D video compression standard 3D-AVS.It proposed two intra prediction algorithms for HEVC and a method of computing local disparity vector.It contained the following three part research work.In the first part,we proposed the gradient-adaptive intra prediction algorithm based on Markov model.This algorithm combined the Markov model and gradientadaptive algorithm to improve the efficiency of intra prediction.It was integrated into HM16.0 which is the reference software of HEVC.The average coding gain was about 1.0% for luma component and about 1.2% and 1.3% on average for two chroma components respectively.In the second part,we proposed the gradient-adaptive intra prediction algorithm base on PDPC which referred to the future video coding technology.This algorithm made use of the relativity among pixels.Experiment results showed the average coding gain is about 0.2% for luma component and about 0.4% on average for two chroma components compared to JEM-2.0.In part three,we proposed the local disparity vector based on time domain and space domain.It made full advantage of the time domain reference frame and current frame.Experiments were carried out to verify the compression efficiency of the proposed algorithm and it was integrated into RFD8.0 which is the reference software of 3D-AVS.Experiment results showed the average coding gain is about0.12% on average for synthesis view compared to RFD8.0.These three algorithms get better experiment results.Compared to current video coding standards and some relative algorithms,they have better performance.