Shape-preserving mesh decimation within a graph-theoretic framework

A new mesh decimation algorithm is proposed to address the problem of obtaining shape-preserving coarse approximations of a highly detailed 3D surface mesh. The input mesh is segmented into multiple, disjoint sub-meshes to facilitate decimation. Given a sub-mesh, various shape clusters are identified and the vertices in those clusters arc labeled as boundary or interior. Shape is preserved by considering only similar-labeled vertex pairs as candidates for a potential merge. Sub-mesh decimation is realized by merging a vertex pair that minimizes the proposed graph energy based cost function. Results demonstrate the efficiency and efficacy of the approach