Position detection system for mechanical components by binocular vision

WANG Caidong1, DONG Xiangsheng1, CHEN Zhihong2,CHEN Xiangyuan2, LIU Jie1, LIU Chao1

(1.College of Mechanical and Electrical Engineering, Zhengzhou University of Light Industry, Zhengzhou

450002, China; 2.Zhengzhou Kehui Technology Co., Ltd., Zhengzhou 450001, China)

Abstract: Aiming at the practical problem of accurately measuring the position of the parts in the process of mechanical parts handling，a mechanical part pose detection system was designed by the binocular stereo vision measurement system platform (VS220). The improved Canny edge detection algorithm was used to identify the edge contour of the components. The feature extraction and stereo matching of the components feature points were performed by the scaleinvariant feature transform method. The mathematical model of pose detection system was established by stereoscopic threedimensional reconstruction method. By solving the conversion relationship between the coordinate systems, the position and direction of the components placed on the conveyor belt were calculated. The pose recognition algorithm program was written by VC++ to realize the shape and pose recognition of the components. A test platform for the pose test system was built and a test experiment was carried out. The experimental results show that the detection system can quickly and effectively detect the position of the components, and provide technical support for automatic loading and unloading of the components.