Hierarchical Compression of Tetrahedral Meshes Through Clustering and Vector Quantization

Abstract

To visualize the unstructured volumetric and surface data Delaunay triangulation is utilised which results in the formation of tetrahedral meshes. These generated tetrahedral meshes possessing Delaunay topology, facilitates the progressive transmission design and coding. Two or three levels of detail of the same data can be acquired by implementation of hierarchical decimation over these meshes. These data sets are then sent to Encoder for compression. A new scheme for geometry data compression has been devised in this paper. Compression algorithm is composed of three stages. Initially centroids and the frequencies are determined by clustering. Then error vectors are generated by deducting vertex location from centroids. To exploit the statistical dependency of these error vectors vector quantization is employed over them in the third and final stage. The compression scheme, progressive transmission and quality of meshes are scalable.