CSAR Seminar

SPEAKER: Eric Shaffer, UIUC/CSAR

TITLE: Mesh Processing Research at CSAR: Coarsening, Quadrangulation, and Optimization

DATE: Tuesday, December 12, 2006
TIME: 12:00 Noon
PLACE: 2240 DCL
1304 W. Springfield Ave., Urbana, IL

ABSTRACT

I will give an overview of some of the current mesh processing research being pursued at CSAR. The talk will focus on mesh coarsening, quadrangulation of triangulated meshes, and mesh optimization. In work sponsored by the Boeing Company, and in conjunction with Michael Garland (Research Scientist at NVIDIA), we are developing a mesh coarsening method that balances the geometric fidelity of the coarsened mesh with element quality. Many popular mesh coarsening algorithms produce approximations containing poorly shaped elements, and so are ill-suited for scientific applications. The coarsening method I will describe employs an error metric that induces an action similar to Laplacian smoothing while generating a geometrically faithful approximation. In addition to showing the results of some coarsening experiments, I will talk about our efforts to improve the algorithm by automating some parameter selection.

I will also describe our work on a method for quadrangulating a triangulated surface mesh. Quadrangulations are the preferred surface representation in many PDE problems, as well as computer graphics operations such as texturing. I will describe the existing quadrangulation algorithm, and discuss our attempts to make the algorithm responsive to surface features and boundaries.

Finally, CSAR has a long-standing effort in the area of mesh optimization, in which mesh element quality is improved via vertex relocation. I will describe some results of applying our mesh optimization software to meshes used in a commercial engineering setting, and provide details about our current efforts to improve the scalability of mesh optimization methods.