Research On Key Technology Of Workpiece Size Measurement System Based On Binocular Vision

As an important part of computer vision,binocular stereo vision aims at getting structure information of three dimensional objects from two dimensional images.The research shows that,comparing with the traditional manual measurement,measuring technology is based on binocular vision,which has the advantages of non-contact,fast measuring,real time and high accuracy.At the same time,it also has developed rapidly and increasingly becomes the most promising technology in solving measure problems of the manufacturing industry.This thesis uses step shaft as the research object.Also,in this thesis,some key techniques problems of binocular vision measurement system were studied,including camera calibration,edge detection,stereo matching and so on.First of all,on the basis of analysis of the advantages and disadvantages of various methods of camera calibration,and combining with the existing laboratory conditions,an improved camera calibration method which based on the Zhang's plane calibration method is proposed.In order to let the detected corner can be closer to the physical position,when doing the initial calibration,this thesis adopts a sub-pixel corner detection method,which based on the Harris algorithm.After obtaining the internal and external parameters of the camera,this thesis uses the nonlinear optimization to improve the calibration precision of the camera;finally,the calibration of the stereo vision system is achieved by using the rotate matrix and translate vector of the single camera.This experimental results show that this proposed calibration method is flexible and can also meet the precision requirement.Secondly,in the respect of edge detection of the workpiece image,after analyzing the advantages and disadvantages of some common edge detection algorithms,this thesis proposes a new improved Canny algorithm on the basis of traditional Canny image edge detection.This thesis validates the improved algorithm through two sets of experiments,and the result shows that the new edge detection method can effectively detect the image edge and lay a good foundation for the later measurement of the dimension of the workpiece.Then,stereo matching is the key link of binocular vision measurement system.In order to achieve accurate measurement of the dimension of workpiece,this thesis needs to use matching algorithm which has high precision and high accuracy rate.Also,this thesis proposes the stereo matching algorithm based on SIFT,after getting the task characteristics,it adopts RANSAC algorithm,at the same time,it excludes the mismatching points and remains the good matches,to further improve the matching accuracy.Finally,the hardware platforms for the binocular vision system is completed and develop measurement of the binocular stereovision software system.This paper made some research for the workpiece dimensional measurement provided an effective method,which is of the theoretical and practical significance.