Octree-Based Progressive Geometry Coding of Point Clouds

Abstract

We propose a generic point cloud encoder that compresses geometry data including positions and normals of point samples corresponding to 3D objects with arbitrary topology. In this work, the coding process is led by an iterative octree cell subdivision of the object space. At each level of subdivision, positions of point samples are approximated by the geometry centers of all tree-front cells while normals are approximated by their statistical average within each of the tree-front cells. With this framework, we employ attribute-dependent encoding techniques to exploit different characteristics of various attributes. As a result, significant improvement in the rate-distortion (R-D) performance has been obtained with respect to the prior art. Furthermore, the proposed point cloud encoder can be potentially used for lossless geometry coding of 3D point clouds, given sufficient levels of octree expansion and normal space partitioning.