The Circle Pure Rolling Method for Point Cloud Boundary Extraction

We introduce a circle rolling method (CRM) for boundary extraction from 2D point clouds. The core idea is to create a circle that performs pure rolling on the perimeter of the point cloud to obtain the boundary. For a 3D point cloud, a plane adsorbs points on both sides to create a 2D point cloud, and the CRM is used to extract the boundary points and map them back into space to obtain 3D boundary points. Continuously moving this plane can obtain a complete boundary, which is called the moving adsorption rolling method (MARM). In this paper, we solve the interference problems in our method caused by unidirectional overlapping points and porous structures and successfully validate the solutions in practical examples. Our point cloud boundary extraction method is faster in 2D and better for surface concavities extracted in 3D compared to existing methods, and it is unaffected by sparse points within the point cloud.