Noise removal algorithm based on point cloud classification

The point cloud data obtained by 3D scanning equipment inevitably has noise. To remove the noise, we propose a point cloud denoising method. In this experiment, according to the characteristics of noise, noise is divided into two types: large-scale and small-scale. Then to remove large scale noise, we use statistical and radius filtering. For smallscale noise dividing point cloud into patches and each patch is fitted into a surface, thus converted into similarity cloud learning and the removal of surface noise is formulated as a convex optimization problem. Experimental results show that the proposed method is superior to the competing method.