Title: Fast self-repairing region growing surface reconstruction algorithm for unorganised point cloud data
Authors: Nan Luo; Quan Wang
Addresses: Computer School, Xidian University, No. 2 South Taibai Rd, Xi'an, China ' Computer School, Xidian University, No. 2 South Taibai Rd, Xi'an, China
Abstract: This paper proposes a fast self-repairing projection-based surface reconstruction algorithm for both closed-form and free-form data sets in region-growing principle, which generates triangles between reference point and its neighbours according to their status and positions on the tangent plane. In this work, the whole framework of meshing and related concepts is outlined first, then the triangulating procedure is summarised to seven different cases according to the status and position of points, and the corresponding triangulating details are presented. To eliminate the potential missing triangles derived from the triangulating procedure, a data structure Single Edge Index Table (SEIT) is developed to track all the boundaries of the generated triangular mesh and dynamically updated along with the triangle forming. After the triangulation, a quick depth traverse on SEIT is carried out to detect all the local holes in reconstructed mesh, followed by a hole filling procedure to wipe off the holes within predefined size. Experiments validate that proposed algorithm reconstructs better mesh for both closed-form and free-form point clouds, and achieves high efficiency.
Keywords: surface reconstruction; region growing; self-repairing; hole filling; single edge index table; triangulating.
International Journal of Computer Applications in Technology, 2017 Vol.56 No.2, pp.121 - 131
Received: 15 Jul 2016
Accepted: 22 Nov 2016
Published online: 03 Oct 2017 *