Fast VVC Intra Coding Algorithm Based on Luminance-Chrominance Feature and Deep Learning

作者：Shicheng Xu, Lei Chen, Chenyang Ding, Jiayi Xu · 年份：2026 · DOI：10.1109/cnml68938.2026.11452472 · 研究领域：Video Coding and Compression Technologies、Image and Video Quality Assessment、Advanced Data Compression Techniques

To address the issues of high intra coding complexity of the new-generation video coding standard Versatile Video Coding (VVC) and the neglect of chroma components' value in existing research, this paper proposes a deep learning-based video coding optimization algorithm utilizing luma and chroma features. The algorithm aims to reduce the intra prediction mode decision overhead of VVC while improving prediction accuracy. Firstly, we constructs a large-scale structured Coding Unit (CU)-level dataset. Secondly, we designs a multi-channel feature fusion strategy, builds a model based on neural networks, and comparatively analyzes the performance improvement of intra mode prediction brought by the joint representation of luma (Y), blue-difference (U), and red-difference (V) components. Experimental results show that after introducing chrominance components, the model's intra-frame mode prediction accuracy on Class A and Class B video sequences is improved by approximately 5%. This algorithm can balance the complexity and performance of VVC coding, providing technical support for efficient coding in scenarios such as ultra-high definition video and real-time video communication.