Surface reconstruction from unorganized point clouds based on edge growing

Abstract

Owing to unorganized point cloud data, unexpected triangles, such as holes and slits, may be generated during mesh surface reconstruction. To solve this problem, a mesh surface reconstruction method based on edge growing from unorganized point clouds is proposed. The method first constructs an octree structure for unorganized point cloud data, and determines the k-nearest neighbor for each point. Subsequently, the method searches for flat areas in the point clouds to be used as the initial mesh edge growth regions, to avoid incorrect reconstruction of the mesh surface owing to the growth of initial sharp areas. Finally, the optimal mesh surface is obtained by controlling the mesh edge growing based on compulsive restriction and comprehensive optimization criteria. The experimental results of mesh surface reconstruction show that the method is feasible and shows high reconstruction performance without introducing holes or slits in the reconstructed mesh surface.