International Journal of Quantum Technologies

Mesh Denoising

Abstract

Constantin Vaillant-Tenzer

In this paper, we study four mesh denoising methods: linear filtering, a heat diffusion method, Sobolev regularization, and, to a lesser extent, the Sinkhorn algorithm. We show that, for an image, the use of the Gibbs kernel is counterproductive. We demonstrate that while Sobolev regularization is the fastest method, it produces the least faithful denoised mesh compared to the original. We empirically show that for large meshes, the heat diffusion method is slower and less effective than filtering, which is not the case for small meshes. Finally, we observe that the first three methods perform significantly better with high-dimensional meshes.

PDF

Journal key Highlights