IGA-based point cloud fitting using B-spline surfaces for reverse engineering

Abstract

Reverse engineering is a viable method to create a 3D virtual model of real physical parts. Usually, reverse engineering consists of two main steps: (1) measure the object and (2) reconstruct it as a 3D model. The measured data are usually represented as a point cloud without topological information and must therefore often be converted into a tensor product B-spline surface format, which has become an industry standard in computer graphics and in CAD systems. In this paper, a new immune genetic algorithm (IGA) for point cloud fitting that fits a noisy 3D point cloud using a B-spline surface with approximate G1 continuity is presented. The point cloud is first segmented into a set of quadrilateral patches. For every patch, a B-spline surface is reconstructed using a least-squares approximation method, and then the surface is optimized to increase the approximation precision using an IGA-based knots adjustment algorithm. Finally, the B-spline patches are stitched together with approximate G1 continuity with a numerical method and the particle swarm optimization (PSO) algorithm. A set of experimental results shows that the proposed method achieves better approximation accuracy than the Bezier-based method and the GA-based method.