Fast CU partition strategy based on texture and neighboring partition information for Versatile Video Coding Intra Coding

The next generation video coding standard, H.266/Versatile Video Coding (VVC), was released by the Joint Video Exploration Team (JVET) in July 2020. Unlike the previous generation standard H.265/High Efficiency Video Coding (HEVC), VVC introduces a quad-tree coding unit (CU) structure with nested multitype trees (QTMT) and expands the internal prediction patterns from 35 to 67. Although these next-generation techniques improve compression efficiency, the computational complexity in internal coding also increases greatly, leading to a significant increase in coding time. To solve this problem, this paper proposes a fast CU partitioning strategy for VVC internal coding based on texture and neighboring partition information. The main contributions of the proposed algorithm are: 1) we fully explore the texture information and adjacent CU partition information to propose a fast method for CU block partitioning, and the proposed method reduces the computational complexity without significantly degrading the coding performance; 2) by exploring the correlation between texture information and CU partitioning patterns, we design a corresponding fast decision algorithm for different sizes of CUs; 3) based on the correlation between the current CU and correlation between neighboring domains, we guide the fast CU division by combining texture information with the division information of neighboring CUs together. The experimental results show that the coding time of this algorithm is reduced by 44.22% compared with VVC, while BDBR is increased by only 0.93% and BDPSNR is reduced by only 0.04 dB.