Research On Fast Intra Coding Method Based On VVC

With the rapid development of the video market,global video traffic is booming,and video compression technology is facing tremendous challenges.Against this background,the International Organization for Standardization has successively proposed video coding standards with higher compression efficiency to meet compression and transmission needs.Compared with the previous generation video coding standard high-efficiency video coding(HEVC),the latest versatile video coding(VVC)can achieve higher compression performance and is applicable to more application scenarios.The mode selection process of video coding is usually designed according to the principle of rate-distortion optimization(RDO).Among many candidate modes,the optimal encoding mode is determined by comparing the rate-distortion cost of each mode.Although this design method can obtain the optimal video quality under the condition of limited bit rate,it also brings higher computational complexity,which poses great challenges for the practical application of video coding technology.Based on the research and analysis of VVC intra-frame mode selection process,this paper proposes corresponding optimization algorithms for the high coding complexity problem in the coarse selection stage and fine selection stage of mode selection.The main work is as follows:(1)A fast intra-frame mode selection algorithm based on gradient is proposed for the problem of high computational complexity in the coarse selection stage of intra-frame mode selection.Firstly,the statistical characteristics of the distribution of the best intra-frame prediction modes are analyzed.Secondly,according to the texture characteristics of the coding block,the prediction modes are divided into four subsets of modes in different directions.Finally,the mapping relationship between the texture direction of the image block and the intra-frame prediction mode set is established by using the Sobel operator to reduce the number of prediction modes in the coarse selection stage and reduce the time complexity.The experimental results show that under the All Intra(AI)coding configuration mode,the intra-frame coding time can be reduced by 20.73%,the bit rate consumption only increases by 0.46%,and the BDPSNR only decreases by 0.03 d B.(2)A fast rate estimation algorithm based on statistical modeling is proposed for the problem of high time complexity in obtaining the rate cost in the fine selection stage of intra-frame mode selection.Firstly,the algorithm fully considers the quantization behavior of dependent quantization(DQ)and the context dependency in entropy coding,and proposes a rate feature that can accurately characterize the context state transition during the encoding process to preliminarily estimate the rate of some syntax elements in the transform unit(TU).Secondly,based on the coefficient distribution characteristics,the coefficient confusion degree feature and the sparsity feature are defined to distinguish the rate impact caused by coefficient distribution differences and to build a TU-level rate model.Finally,the algorithm separates large TU and small TU for modeling according to the rate composition characteristics to achieve more accurate rate estimation.By statistically analyzing and regression training video sequence samples,a relatively accurate fast rate estimation model is obtained.The experimental results show that under the Random Access(RA)coding configuration mode,the proposed fast algorithm can save 16.29% of the RDO time while only increasing the bit rate by 1.57%.The experimental results of the fast intra-frame encoding joint algorithm show that,in the AI encoding configuration mode,the proposed algorithm reduces the intra-frame encoding complexity by 54.11% without significant loss in video quality,and the bit rate only increases by 1.66%.