Mesh compression: Theory and practice.

Three-dimensional meshes (3D meshes, for short) are fast becoming an emerging media type, used in a variety of application domains such as engineering design, manufacture, architecture, bio-informatics, medicine, entertainment, commerce, science, defense, etc. The volume of data of this media type that is being circulated on the internet is increasing very rapidly and is being used as frequently as other media types like text, audio (1D), images and video (2D). Hence, 3D meshes need good processing and visualization methods. Also, the sizes of these meshes are much greater than the other media types mentioned above and often exceeds the memory and bandwidth available for their storage and transmission. Compression schemes for such large 3D meshes have become a subject of intense study lately. Meshes are either made up of triangles or quadrilaterals. Meshes made up of only triangles are called triangle meshes and meshes made up of quadrilaterals are called quadrilateral meshes (quad meshes, for short). A mesh is described by specifying its geometry (vertex coordinates) and its connectivity (adjacencies of the triangles or quadrilaterals). Previous research on mesh compression has been mostly for triangle meshes. Quad meshes were traditionally handled by first triangulating them and then applying triangle mesh compression techniques. In order to avoid this additional triangulation step, a direct technique is proposed for compressing and decompressing the connectivity of quad meshes. This technique takes a quad mesh as input and encodes its connectivity as a sequence of opcodes which can be restored back to the quad mesh, using the decompression technique. A data structure called EdgeTable is introduced to aid in the traversal of a quad mesh during compression. Also, a technique based on constrained Delaunay triangulation for reconstructing the connectivity of a 2D mesh from its geometry and a minimum set of edges is proposed. Source: Masters Abstracts International, Volume: 44-03, page: 1393. Thesis (M.Sc.)--University of Windsor (Canada), 2005