A Heuristic Approach to Reconstruct Triangle Mesh from Unorganized Point Cloud

A heuristic algorithm was presented to reconstruct triangle mesh from unorganized point cloud, and a new error evaluation rule was presented to measure the reconstruction errors. The triangle meshes were reconstructed through two phases of mesh growing based on different strategies, where regions with high point density usually grow at the first phase and the remaining regions grow later. In each phase of mesh growing, the smoothest regions always grow firstly and then largely avoid errors emerging in sharp regions. The reformative best point selection method can improve the topology and geometry quality of the reconstructed meshes. The presented error estimation was inspired by the level of detail model and was also can be used in other region grow reconstruction approaches. Experiments show that the algorithm is efficient and effective.