Research On Measurement Method Of Artifact Size Based On Binocular Vision

With development of computer technology, image processing based on machine vision has been more popular in object measurement in recent years. Under the background of “Industry 4.0", machine vision technology is very important. As a branch of the machine vision, the binocular stereo vision technology can not be ignored. It is that extracting and anglicizing information from two-dimensional image and understanding of space objects.In production of advanced manufacturing, it can improve manufacturing efficiency and quality and reduce cost using advanced online testing technology. In this paper, the measurement system of artifact size based on binocular vision is studied.(1) In the camera calibration, in order to make the binocular vision measurement system to achieve rapid and accurate measurement, it is necessary to choose a simple and convenient, high precision, easy to achieve calibration method. Based on the existing camera calibration methods, a camera calibration method based on Zhang is proposed.(2) In the image denoise, for there is highlight in smooth surface of the artifact in the detection, a method is proposed which based on color space conversion technology. It combines brightness adjustment with polynomial method. It not only reduces the brightness value of the high brightness region, but also does not change the brightness value of the low intensity region. And the original image texture feature information can be well retained.(3) In the stereo matching, the binocular vision measurement system requires the precision, accuracy rate and efficiency of stereo matching algorithm are high. A fast stereo matching algorithm based on SIFT feature is proposed. After getting stable SIFT feature,the transform matrix is solved and refined with RANSAC algorithm. The matching speed and accuracy are improved by combining guided matching method with voting filter method.(4) Based on the machine vision measurement research, the binocular vision system of artifact size measurement system is designed and developed. The platform of binocular vision system is completed and software system is debugged. The measurement of artifact size is realized.