Simplification of 3D point clouds sampled from elevation surfaces

This paper introduces a new technique to simplify a 3D point cloud sampled from an elevation surface and organized in voxels. The method consists of three steps: in a first step, the boundary of the surface is extracted and simplified; in a second optional step, we roughly simplify the surface inside its boundary; in a third step, we present an elaborate method for simplification while keeping its boundary. Our method preserves the distribution of points, the initial geometry and characteristics of the surface, even with high simplification rates.