Optimization of 3D GIS model reconstruction algorithm based on laser point clouds

Abstract

As the data basis of 3D geographic information service, 3D model is widely used in 3D terrain visualization, smart city, geological monitoring and other fields. 3D laser point clouds have become one of the main data sources for 3D GIS modeling due to their high precision and fast modeling. When researchers mesh laser point clouds and generate some elongated triangles due to the rule constraints of the algorithm itself, elongated triangles can cause the burden of mesh optimization, affecting the efficiency of mesh reconstruction. In this paper, based on the incremental insertion algorithm, we mitigate the adverse effects of elongated triangles by constructing a non-fully partitioned KD tree index and introducing control points to influence the insertion order of point clouds. We use the proposed algorithm, BRIO and KD tree algorithms to reconstruct Bunny and Horse. We found that the proposed algorithm is superior to the other two algorithms in time efficiency and can suppress the appearance of elongated triangles well.