The computation of delaunay triangulation of LiDAR point cloud based on GPU

Abstract

The algorithms of construction of Delaunay triangulation are relatively mature, but when the data volume is very large, the traditional algorithms which are based on CPU are low in efficiency. This paper adopts CUDA of NVIDIA in parallel computing to construct Delaunay triangulation. In this paper, the PBA algorithm is adopted to perform the EDT to get a raster Voronoi diagram. The final Delaunay triangulation network is transformed from the raster Voronoi diagram. At last, the paper use OpenGL API to realize the 3D rendering of the Delaunay triangulation network.