Measurement Algorithm of Mounting Holes Based on Binocular Stereo Vision

Considering the high reflectivity of metal components and the difficulty of binocular stereo matching, to obtain the diameter of the mounting hole accurately, this paper presents a effective measurement scheme, based on epipolar-line constraint algorithm for binocular stereo vision and the theory of direct least square method. First of all, using Canny edge detection operator to extract the sub-pixel contours of mounting holes, which can be obtained from the gray image capture of left and right camera; with Turkey weight function fitting ellipse, reducing the influence of outliers to get a better fitting precision; The line laser, which is used for searching for feature points, moves in a fixed direction and scans in the field, in order to determine the center of the laser strip accurately. This article, combining extreme value method and gray-gravity method, shows a strategy, which can determine the effective range of computing automatically. Based on the geometric characteristics of the circular space, we can get the radius of mounting holes and the coordinates of the center point of the circular hole. It proves that the method can be implemented with a strong and high accuracy by multi group experiments.