Three-Dimensional Real Object Modeling Based on Modified Closest-Point Algorithm for Monitoring in Intelligent Manufacturing

This article presents a modified iterative closest point algorithm (ICP) for three-dimensional modeling to monitor an intelligent manufacturing. This modified ICP allows comparing the created object with a benchmark object for the detection of defects. The proposed algorithm is based on bearing angle images, global control points, and Manhattan metric. This algorithm eliminates two shortcomings: settings of the initial rotation matrix and the shift vector, and the time-consuming finding of closest points. The experiments show that the proposed algorithm insignificantly loses its alignment accuracy to the basic and generalized ICP-algorithms with the 100% overlap region between the clouds. However, the proposed algorithm demonstrates a higher alignment accuracy when the overlap region decreases. Besides, the algorithm wins in run-time compared with the basic and generalized ICP algorithms.