A Three Dimensional Mesh Improvement Algorithm Based on Curvature Flow

In order to suppress undesirable noises in triangular meshes for quality improvement, an effective smoothing method based on curvature flow is proposed in this paper. For triangular meshes generated from planar contours, the continuity information amongst adjacent contours is utilized to determine whether a vertex represents a geometric feature to preserve or an isolated noise to smooth. Before smoothing the original mesh with an isotropic diffusion process, local adaptive refinement is performed where the sharp features are detected. In doing so, the smoothing method has both the benefits of the isotropic diffusion method, simplicity and stability, and that of the anisotropic diffusion method, feature preserving. The presented method is implemented along with several other smoothing methods, and experiments that smooth triangular meshes reconstructed from contours are performed. Experimental results demonstrate that our method can smooth triangular meshes in an effective way.