Research On Video Intra Coding And Rate Distortion Optimization Technologies

With the availability of several new kinds of videos and applications,such as VR,360-degree video and live video etc.,human beings are keen to experience the display video with good quality.Therefore,how to efficiently compress the huge amount of video data is a critical issue in video compression area.Addressing this problem,JVET(Joint Video Exploration Team)established recently plans to develop the next generation of video compression standard and has published a video coding reference software exploration model called JEM(Joint Exploration Model).JEM still applies the hybrid video coding framework as in HEVC(High Efficiency Video Coding)and incorporates several new coding techniques to increase coding efficiency.Meanwhile,the second generation of national video compression standard AVS-2(Advanced Audio and Video Coding Standard)was published and the next generation of national video compression standard AVS-3 will be started recently.As for the new video coding techniques applied into JEM and AVS-2,more explorations and research work are needed to further improve the coding performance.This paper focuses on this problem and the contributions are as follows.1.Research on Video Intra Coding techniques(1)Commutative mode-dependent transform for video Intra codingIn order to further improve the Intra coding performance,this paper proposes a commutative mode-dependent transform method by studying the statistical properties of the residuals of both luma and chroma components respectively.In our proposed method,transform cores of the specified Intra prediction mode for luma and chroma components are determined by the statistical properties of the residuals.Specifically,transform cores of two adjacently directional Intra prediction modes will be exchanged.(2)Multi-lines based position-dependent Intra prediction modeBased on one line Intra prediction,this paper proposes a multi-lines based position-dependent Intra prediction mode which utilizes the reconstructed pixels in multiple lines next to the current coding unit.2.Research on rate distortion optimization techniques(1)Adaptive GoP quantization parameter offset adjustment methodThis paper focuses on the compatibility and scalability problems of previous table based Sub-GoP QP offset adjustment in the random access coding method.An enhanced GoP level QP offset adaptation method is proposed.(2)Optimization for GoP level QP offset under RA configuration with GoP size 16Since AVS-2 codec extends the GoP size from 8 to 16 under RA configuration,this paper proposed a modified GoP level QP offset adjustment method under RA configuration with GoP-16.(3)Layer based Temporal Dependent Rate-Distortion OptimizationThis paper incorporates TD-RDO into AVS-2 codec and proposes a novel temporal-layer QP adaptation strategy,which can show a better performance with negligible coding complexity increase.