Detection Of Spacial Circular Hole Based On Binocular Stereoscopic Vision Technology

The traditional means of detection has been unable to meet the work-piece processing in the online testing with the development of industrial processing automation.Machine vision measurement method has a wide range of applications in aerospace surveying,non-contact measuring range,robot navigation or other fields.I studied the binocular vision measurement model after understanding the actual testing needs of the factory.Simultaneously,I have made clear my work goals and work points after the research on foreign machine vision measurement technology.Finally,I have proved the feasibility of my program through the measurement of round holes.In particular,it is necessary to note that I have made an accurate and efficient way to search for match points between the two images by using the geometric characteristics of the circle on the basis of the criterion of line constraint.Moreover,the accuracy of the method is proved by experiments.This subject mainly studies the binocular vision system model,the corner extraction algorithm,the edge detection algorithm,the feature point matching technology,the space circle fitting technique.The main contents are shown below:(1)The binocular stereoscopic vision system model is set up,and the four coordinate systems in the field of machine vision are introduced one by one.Then,I deduced the transformation between the various coordinate system.(2)I used a more accurate sub-pixel corner detection method based on the traditional Harris corner detection algorithm;I have thoroughly studied the Zhang's calibration method used in this project.(3)I analyzed the Canny edge detection algorithm emphatically.Meanwhile,the superiority of Canny algorithm to circular edge detection is proved by experiments.I carried out a SIFT matching experiment on the target circular hole.Combining with the task of detecting the geometrical characteristics of the target hole,I propose a matching algorithm that is suitable for finding matching points on the edge of a circle.(4)Through experiments,I realized the size inspection of the hole of the work-piece,I compare the test result with the actual size of the work-piece hole,and carry out the program evaluation and error analysis,and put forward the improvement plan and research prospect.In this thesis,Da Heng Industrial Camera DH-HV1351UC-ML and the industrial lens are used as the image acquisition hardware devices.The corner detection,camera calibration and ellipse fitting are realized by Matlab.The programming feature matching is realized based on Matlab programming experiment.