Fast Decision Making of Point Cloud Video Geometry CU Partition Based on Occupied Pixels

The video-based point cloud compression Standard (V-PCC) uses file-based projection technology to map point cloud data in three-dimensional space to two-dimensional space, creating geometric graphs and attribute graphs, then, with the help of the currently very mature two-dimensional video image coding technologies, such as high-efficiency video coding (HEVC), Versatile Video Coding (VVC), etc., these generated images are encoded and processed. Compared with natural video, point cloud video has a larger amount of data, and V-PCC coding improves coding efficiency but is more complex. In order to achieve effective complexity reduction and improve coding performance in the encoding process, this paper proposes a fast decision making method of coding unit (CU) partition of point cloud video geometry based on occupied pixels. In this method, the occupancy information is represented by block based occupancy map, and the early termination partition of unoccupied blocks in geometric graph is realized by generating occupied pixel marker block. Then, an unsupervised hierarchical clustering method is used for different scales of fully occupied blocks and partial occupied blocks containing occupied pixels, so as to realize the fast decision of optimal CU partition. The experimental results show that when the proposed method is used to process different dynamic point clouds in All Intra (AI) configuration, the average time of geometric graph obtained after coding is saved by 62.76%, while the BD-TotalRate only increases by 0.04% and 0.10%. Compared with other methods, the coding complexity is reduced. A good balance is also obtained for coding losses.