CS448B Critique: Geometric Compression Through Topological Surgery

This paper details a geometric compression technique for triangulated 3D polyhedral models which significantly reduces their storage and transmission costs. The method is based on entropy and run-length encoding of vertex and triangle spanning trees. This losslessly compresses mesh connectivity to less than two bits per triangle on average. Vertex positions and properties are quantized and encoded by entropy and predictive coding based on ancestors in the vertex tree.