A novel algorithm for segmentation of complex point cloud with 3D elastic model

Segmentation of 3D object from complex point cloud is usually performed in 2D space by using 2D algorithms. This paper presents a novel 3D algorithm for segmentation of complex point cloud. A closed B-spline surface is selected as 3D elastic model. After initialization, control points of B-spline surface are attracted by a 3D force field and move to the surfaces of target object. 3D GVF derived from the complex point cloud acts as the force field. Finally, the deformable B-spline surface can enclose the target surfaces. Points belonging to target object are segmented and mini-defects are repaired. Experimental results show the efficiency of the proposed 3D algorithm.