Fast CU Partitioning Algorithm Based on Decision Trees for Virtual Reality 360 Videos.
International journal of pattern recognition and artificial intelligence(2020)
摘要
For the virtual reality 360[Formula: see text] videos, equirectangular projection (ERP) is a commonly used projection format. However, its high resolution brings extraordinary huge computational complexity in encoding. In order to speed up the intra coding process, a fast coding unit (CU) partitioning algorithm based on regional decision tree is proposed in this paper. The frame image is divided into two regions from a statistical point of view, and the earlysplit and pruned decision trees are established using light weight sample attributes for each region. With the help of these decision trees, the CU partitioning process is accelerated. Compared with the original algorithm of HM16.20, the proposed algorithm can reduce the encoding time by 28%, while BD-rate only increases by 0.27%.
更多查看译文
关键词
CU partitioning,decision tree,360 degrees videos,HEVC
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要