Video-Based Point Cloud Compression Using Density-Based Variable Size Hexahedrons

Abstract

The most advanced standard for compressing dynamic point clouds, a recently created media format, is video-based point cloud compression (V-PCC). V-PCC relies on creating patches and converting 3D input into 2D frames to apply existing video coding. However, creating patches according to the normals and packing irregular projected patches produce many unoccupied pixels in 2D frames, affecting the temporal prediction and inefficiency of 2D video coding. In addition, unoccupied pixels increase the size of 2D frames, accordingly, the required bitrate. This paper introduces variable-sized hexahedrons segmentation as an alternative to patch creation to reduce the number of unoccupied pixels in the 2D frames. Furthermore, different areas of a point cloud are treated differently according to the density of those areas resulting in capturing points more accurately. This paper investigates combinations of different hexahedrons' sizes, and the experimental results demonstrate that the proposed method with the appropriate hexahedrons' sizes outperforms V-PCC.