Spectral Surface Reconstruction

Surface reconstruction recovers a watertight surface mesh from a given set of oriented or unoriented point clouds. Given a tensor field that represents the orientation of a point set, a spectral reconstruction process then computes the implicit function by solving an eigenvalue problem that aligns its gradient with the normal vector of point clouds. In this project, we will focus on improving the spectral reconstruction quality in the aspect of discretization and numerical robustness. The final delivery will contribute to a part of the CGAL surface reconstruction component.