Point Cloud Reconstruction From Truncated Geometry-Based Streams

• DOI: 10.36227/techrxiv.14720991.v1
• Publication Date: 2021-06-03T15:25:51Z
• Authors: DIOGO GARCIA; Andre Souto; Gustavo Sandri; Tomas Borges; Ricardo Queiroz
• Keywords: Octree; Bitstream

Geometry-based point cloud compression (G-PCC) has been rapidly evolving in the context of international standards. Despite inherent scalability octree-based geometry description, current G-PCC attribute techniques prevent full for compressed clouds. In this paper, we present a solution to add attributes using region-adaptive hierarchical transform (RAHT), enabling reconstruction only portion original bitstream. Without information, one cannot compute weights which RAHT relies on calculate its coefficients further levels detail. order overcome problem, propose linear relationship approximation relating downsampled truncated inverse at that same level. The parameters are sent as side information. After truncating bitstream corresponding given octree level, can, then, recreate Tests were carried out and results attest good quality proposed technique.