Abstract
In order to remove noise on a mesh, smoothing techniques have been developed. We have proposed the smoothing method with the discrete Laplacian implemented in the diffusion process. For the feature preservation, the target model is segmented by the detected feature lines in our method. However, such treatment will lead to the problem that user must deal with many feature lines for a huge complex model. In this paper, we adopt the anisotropic diffusion flow for more efficient feature preservation. Simulation results show that the presented method worked well on a closed and an open model with preserving its features.
