In the image-guided automated assembly of circular hole metal workpieces,the image of the round hole metal workp iece has weak or even no texture,resulting in low positioning accuracy of the round hole pose.In order to solve the problem,a parallel binocular structure was constructed by an industrial robot with a monocular industrial camera moving horizontally to obtain binocular images.The basic matrix of the image was obtained by rough matching the feature points extracted from the contour of the circular hole in the binocular image,and then the contour of the circular hole in the binocular image was extracted by the arc-based ellipse detection.Finally,the above information was used to perform binocular precise stereo matching based on epipolar constraints and the three-dimensional coordinates and poses of the relevant points were obtained.The experimental results show that the accuracy of the distance between the hole position and the hole center is maintained within 1 mm,the accuracy of the position of hole is maintained within 0.3°.