Optimal Area Polygonization by Triangulation and Visibility Search

Abstract

The aim of the “CG:SHOP Challenge 2019” was to generate optimal area polygonizations of a planar point set. We describe here the algorithm that won the challenge. It is a two-phase algorithm based on the node-insertion move technique, which comes from the TSP. In the first phase, we use constrained triangulations to check efficiently the simplicity of the generated polygonizations. In the second phase, we perform visibility searches to be able to generate a wider variety of polygonizations. In both phases, the simulated annealing metaheuristic is implemented to approach the optimum.