Research On Optimization Of Cross-component Linear Model Prediction In VVC

The new generation of video coding standard,Versatile Video Coding(VVC)has been finalized in July 2020.Compared with High Efficiency Video Coding(HEVC),VVC has added a lot of new technologies.Among them,the introduction of cross component linear model(CCLM)prediction technology uses the correlation between components to improve the prediction efficiency of chroma components.However,this technology significantly increases the computational complexity of the chroma prediction mode decision-making process,and there are problems such as low prediction accuracy for coding blocks with complex texture.This thesis conducts research on the above issues,and the main content and innovation are as follows:(1)Aiming at the problem of increased computational complexity in the decision process of chrominance prediction mode due to the introduction of CCLM technology,a fast decision algorithm for intra-frame chrominance prediction candidate modes based on texture features is proposed.Through statistical analysis,we found that the choice of intra-frame chroma prediction candidate mode is closely related to the texture complexity of the coding unit(CU).In addition,the choice of direct mode(DM)is closely related to the texture similarity between the current chroma CU and its Corresponding brightness CU.An improved Sum of Modulus of Gray Difference(SMD)to measure the texture complexity of CU is designed,and is used to to disable irrelevant candidate modes.At the same time,the structural similarity index(SSIM)is used to decide whether the DM mode is the optimal mode in advance.The experimental results show that compared with the reference model VTM8.0,the proposed algorithm can reduce the coding time by 12.92% on average,and increases the BD-rate of Y,U,and V components by only 0.05%,0.32%,and 0.29% respectively.(2)The CCLM prediction mode only relies on a limited number of adjacent samples to calculate the parameters of the linear model,which degrade the prediction performance for blocks with complex texture.A cross-component linear model optimization algorithm is proposed for joint angle prediction,which uses the spatial correlation and inter-component correlation of the current chroma CU,and uses the average of the predicted values of the angle mode and the CCLM mode to predict the chroma CU,so as to improve the prediction accuracy.The experimental results show that compared with the reference model VTM8.0,the algorithm has a 0.02%,0.65%and 0.80% decrease in the BD-rate of the Y component,U component and V component,respectively,which improves the coding efficiency of the original algorithm.And the encoding time has hardly changed.