New Mesh Signal Processing Algorithms

Monday, May 14, 2001 - 11:00am - 12:00pm
Keller 3-180
Several closely related methods have been proposed in recent years to smooth, denoise, edit, compress, transmit, and animate very large polygonal models, based on signal processing techniques, constrained energy minimization, and the solution of diffusion differential equations. A number of these have been proposed to fix some of the drawbacks of the simple Laplacian smoothing algorithm, such as shrinkage, tangencial drift, and ridge over-smoothing. Almost none of these extensions beat Laplacian smoothing in simplicity and ease of implementation. In this talk I will describe new signal processing algorithms, all based on various extension and modifications of the discrete Laplacian operator defined on a polygonal mesh. These new algorithms, solve some of the existing problems while sharing the simplicity of Laplacian smoothing.