Abstract:In order to solve the problem of lack of texture in welding image, a stereo matching algorithm for weak texture detection by cross-scale polymerization is studied. According to Zhang’s calibration method, by calibrating binocular vision sensor, the external and internal parameters of binocular vision sensor is obtained. Then, a series of images, such as median filtering, binary image processing are preprocessed. Because stereo matching is the most critical link of binocular vision technology, the algorithm of cross-scale polymerization of weak texture detection is used to obtain parallax map of weld image by stereo matching of weld image. The distance can be recovered from the parallax to obtain the height information of the welding gun. The welding seam images collected by binocular vision sensor is tested and the results show that the reliability of the proposed algorithm is verified.