Abstract:
Mesh Compression is a broad research area with applications in a lot of different areas, such as the handling of very large models, the exchange of three dimensional content over the internet, electronic commerce, the flexible representation of volumetric data and so on. In this thesis the mesh compression method of the Cut-Border Machine is described. The Cut-Border Machine encodes meshes by growing a region through the mesh and encoding the way, in which the mesh elements are incorporated into the growing region. The Cut-Border Machine can be applied to triangular and tetrahedral meshes. Although the method is not too complicated, it achieves very good compression rates. In the tetrahedral case the Cut-Border Machine performs best among all known methods. The simple nature of the Cut-Border Machine allows on the one hand for a hardware implementation and performs also as software implementation extremely well. On the other hand the simplicity allows for a theoretical analysis of the Cut-Border Machine.
It could be shown, that for planar triangulations a slightly modified version of the Cut-Border Machine runs in linear time in the number of vertices and that the compressed representation only consumes linear storage space, i.e. no more than five bits per vertex.
Besides the detailed description of the Cut-Border Machine with several improvements and optimizations, the thesis gives an introduction to meshes and appropriate data structures, develops several coding techniques useful for mesh compression and gives a broad overview of related work. Furthermore the author improves the encoding efficiency of several other compression techniques. In particular could the algorithmically achieved upper bound for the encoding of planar triangulations be improved to ten percent above the theoretical limit, what is the best known result up to now.