Rotational-guided optimal cutting-plane extraction from point cloud

Abstract

This paper presents a simple yet powerful rotational-guided optimal cutting-plane extraction method which can provide an effective way for skeleton extraction, simplification, surface reconstruction and boundary extraction. The optimal cutting-planes can be obtained by several steps: calculating the cutting-points and initial cutting-plane, rotating reference cutting-plane on each cutting-point along the given rotation axis to generate a group of cutting-planes, calculating the area of these planes and selecting the plane with the smallest projected area. The primary advantage of the proposed method is that it does neither place strong requirements on the quality of input point cloud nor on the geometry of captured shapes. The experimental results demonstrate that our method can extract the optimal cutting-planes effectively for an unoriented raw point cloud model, even in the presence of noise and missing data.