Simplification of point cloud data based on Gaussian curvature

In order to highlight the characteristic region of 3D models, this paper proposes a point cloud data simplification algorithm based on Gaussian curvature. In order to find the nearest K points, we need to divide the space of point cloud according to 3D cube algorithm firstly. Secondly, we build a curved surface based on normal constraints so as to get the Gaussian curvature of the nearest K points. Then we simplify cloud point based on evaluation function which is derived by Gaussian curvature. At last, we construct 3D model according to the simplified point data. Experiments show that our method can well preserve the point cloud in characteristic area.