CNN-Based Fast CU Partitioning Algorithm for VVC Intra Coding.

Jun Xu,Guoqing Wu,Chen Zhu,Yan Huang,Li Song

DOI: https://doi.org/10.1109/icip46576.2022.9897378

2022-01-01

Abstract:Over a year has passed since the finalization of Versatile Video Coding (H.266/VVC), yet it is still far from practical deployment, a major reason being the excessive complexity. The flexible and sophisticated quad-tree with nested multi-type tree partitioning structure in VVC provides considerable performance gains while bringing about an exponential increase in encoding time. To reduce the coding complexity, this paper proposes a Convolutional Neural Network (CNN) based fast Coding Unit (CU) partitioning algorithm for intra coding, which accelerates CU partition through predicting the partition modes with texture information and terminating redundant modes in advance. Corresponding classifiers are designed for different CU sizes to improve prediction accuracy. Low rate-distortion performance degradation is guaranteed by introducing performance loss due to misclassification into the loss function. Experiments show that the proposed method can save encoding time ranging from 38.39% to 62.33% with 0.92% to 2.36% bit rate increase.