Vision Guided Automatic Measuring in Coordinate Metrology
Yuhong Qin,Lei Wang,Lusheng Xie,Yuanqing Huang
DOI: https://doi.org/10.1117/12.819550
2008-01-01
Abstract:A novel automatically measuring planning method in coordinate metrology based on computer vision is presented in this paper. An active stereo vision system is established by attaching a CCD camera to the mechanical probe of the coordinate measuring machine (CMM). Through the movement of the probe of the CMM, as well as the camera, 3D edge characters of the object can be acquired, which are used as clues for automatically coordinate measuring. A multi-baseline matching method is presented to overcome the ambiguity in stereo matching, and a quadratic interpolating is used in sub pixel matching to get continuous depth image. The matching is only done on character edges in images, so it is much faster and more robust. Two methods of measuring path planning are put forward, in one way, a 2D characteristic edge image which are often stand for rapidly changes in depth or curvature of object surface can be acquired by projecting 3D edge characters to a scanning plane, and then the sampling points of mechanical probe are selected depending on the edge image. In the other way, surface patches are fitted to these 3D edges, and the sampling grid is determined by the type and area of every patch. Using these techniques, a highly automated high-speed, high-precision, 3-D coordinate acquisition system based on multiple-sensor integration can be developed. It has potential applications in manufacturing problems as metrology, inspection, and reverse engineering.