Segmentation-based semi-regular remeshing of 3D models using curvature-adapted subdivision surface fitting

This paper proposes a novel method of semi-regular remeshing for triangulated surfaces to achieve superior triangles lead to advanced visualization of 3D model. It is based on mesh segmentation and subdivision surface fitting which uses curvature-adapted polygon patches. Our contribution lies in building a sophisticated system with three stages, i.e., curvature-aware mesh segmentation, submesh surface fitting to generate a high-quality semi-regular mesh and finally, stitching the segments using an efficient algorithm. Our method uses centroidal Voronoi tessellation and Lloyd’s relaxation to generate curvature-adapted site centers. Geodesic distances from site centers are used for labeling segments and indexing corner vertices for each segment boundary. Using information of site centers and corner vertices, feature-adapted polygonal patches are generated for each segment. These patches are then subdivided and optimized using squared distance metric to adjust position of the subdivision sampling with segment details and prevent oversampling. At last, an efficient stitching algorithm is introduced to connect regular submeshes together and build the final semi-regular mesh. We have demonstrated the results of our semi-regular remeshing algorithm on meshes with different topology and complexity and compared them with known methods. Superior triangle quality with higher aspect ratio together with acceptable distortion error is achieved according to the experimental results.