A vision-aided reverse engineering approach to reconstructing free-form surfaces

A vision-aided reverse engineering approach (VAREA) has been developed to reconstruct least-square free-form surface models from physical models, with a coordinate measurement machine (CMM) equipped with a touch-triggered probe and a computer vision system. The VAREA integrates computer vision, surface data digitization and surface modelling into a single process. Two main steps are applied in this innovative approach. The initial vision-driven surface triangulation process (IVSTP) generates a triangular patch by using stereo image detection and a constrained Delaunay triangulation method. The adaptive model-based digitizing process is then used to refine the surface reconstruction and to control accuracy to within user-specified tolerances. As a result, a least-squares bicubic B-spline surface model with the controlled accuracy of digitization can be obtained for further application in product design and manufacturing processes. More than 85% reduction has been achieved in the time required to construct free-form surfaces using this approach, as compared with traditional manual methods with CMM. Therefore, product design lead time can be significantly reduced.