Feature Edge Extraction Method of Triangle Meshes Based on Tensor Voting Theory

Feature detection and extraction plays important role in mesh editing.However,most existing algorithms often fail in dealing with irregular meshes.To overcome those problems,an algorithm for extracting feature edges of triangle meshes based on tensor voting is presented.First,all vertices of an input mesh are classified according to the observation that there is a close correspondence between the eigenvalue distribution of the tensor voting matrix and geometrical features.The classified vertices are then optimized by connecting breakpoints.Region growing is performed for each seed triangle and the boundaries of the regions are extracted as the edges.The experimental results show that the proposed algorithm is effective in nearly all cases,including models with non-uniformly distributed triangles,long and narrow triangles or even holes.It is also robust on noisy data.