Delaunay-restricted Optimal Triangulation of 3D Polygons

Abstract

Triangulation of 3D polygons is a well studied topic of research. Existing methods for finding triangulations that minimize given metrics (e.g., sum of triangle areas or dihedral angles) run in a costly O(n 4 ) time [BS95, BDE96], while the triangulations are not guaranteed to be free of intersections. To address these limitations, we restrict our search to the space of triangles in the Delaunay tetrahedralization of the polygon. The restriction allows us to reduce the running time down to O(n 2 ) in practice (O(n 3 ) worst case) while guaranteeing that the solutions are intersection free. We demonstrate experimentally that the reduced search space is not overly restricted. In particular, triangulations restricted to this space usually exist for practical inputs, and the optimal triangulation in this space approximates well the optimal triangulation of the polygon. This makes our algorithms a practical solution when working with real world data.