Global mesh optimization with automatic surface structure preservation

We propose a robust mesh optimization algorithm to improve the mesh quality of a 3D model. Our algorithm consists of two parts. First, we automatically extract a desired number of feature points that is sufficiently represented the surface structure of the original model by using a normal-based surface segmentation technique. The extracted feature points are taken as the geometric constraints to efficiently reduce the appearance error produced by the mesh optimization process. In the second part, we combine a simple geometric operation and a global vertex reposition technique to efficiently improve the regularity of mesh connectivity and vertex distribution of optimized mesh. The experimental results demonstrate the robustness of the proposed algorithm.