3D IMAGE MESH ENTROPY CODING

Abstract

In this article we have introduced a compressed representation for triangular meshes. Because proximity in this vertex spanning tree often implies geometric proximity of the corresponding vertices, we can use ancestors in the tree to predict vertex positions, and thus only need to encode the difference between predicted and actual vertex positions. When vertex coordinates are quantized (i.e., truncated to the nearest number in a fixed-point representation scheme), these corrective vectors have on average smaller magnitude than absolute positions and can therefore be encoded with fewer bits. Furthermore, the corrective terms are then compressed by entropy encoding using, for example, Huffman or arithmetic coding as in the JPEG/MPEG standards.