Forward Prediction-Guided Cross-Partition Targeted Pruning for VVenC

With the rapid development of digital video technology, video coding plays a crucial role in multimedia communication, network streaming, and other fields. Efficient video encoding algorithms can significantly reduce data storage and transmission costs, and improve video quality. This article focuses on VVenC, an open-source encoding platform based on the latest video encoding standard H.266/VVC, and studies and optimizes its video encoding block partitioning strategy. In the encoding process, the division of encoding units and pattern prediction are interdependent, and the accuracy of division is transmitted to the accuracy of pattern prediction, and vice versa. In order to ensure encoding efficiency and reduce encoding time, we propose a pruning algorithm based forward prediction-guided and cross-partition (FPG-CTP). We perform targeted pruning on the horizontal and vertical block partition based on the prediction results of the forward prediction model. The experimental results show that the optimization strategy proposed in this paper significantly improves compression performance while keeping the encoding complexity basically unchanged. On various test sequences, compared to the original VVenC encoder, the average time savings reached 48.67%, and the average bit rate increased by 1.81%. In addition, the work of this article also provides reference for optimizing block partitioning in other video coding standards, and has a certain promoting effect on the further development of video coding technology.