Simplification with Feature Preserving for 3D Point Cloud

Point cloud often miss the geometric feature in the process of being simplified. This paper proposes a simplified algorithm which can preserve geometric features of point cloud. Firstly, the point cloud is down-sampled according to its density. Secondly, the average curvature is calculated after point cloud is down-sampled, through which we can get the feature of point cloud. Then, the region growing clustering method is used to get the simplified point cloud according to the curvature threshold. Finally, the feature of point cloud and the simplified points are fused, removing those duplicated points, in order to get the strong featured point cloud. Experiments show that the proposed algorithm can effectively simplify point cloud while preserving the feature of the points.