Efficient Compression of 3D Dynamic Mesh Sequences

Abstract

This paper presents a new compression algorithm for 3D dynamic mesh sequences based on the local principal component analysis (LPCA). The algorithm clusters the vertices into a number of clusters using the local similarity between the trajectories in a coordinate system that is defined in each cluster, and thus transforms the original vertex coordinates into the local coordinate frame of their cluster. This operation leads to a strong clustering behavior of vertices and makes each region invariant to any deformation over time. Then, each cluster is efficiently encoded with the principal component analysis. The appropriate numbers of basis vectors to approximate the clusters are optimally chosen using the bit allocation process. For further compression, quantization and entropy encoding are used. According to the experimental results, the proposed coding scheme provides a significantly improvement in compression ratio over existing coders.