Research On Key Measurement Technology Of Binocular Stereo Vision For Welded Parts

In the manufacturing industry,welding is mostly carried out in a streamlined manner,which has the characteristics of large batches and high welding accuracy requirements for welded parts.With the widespread application of intelligent manufacturing and automated inspection technology,current requirements for welding quality inspection in industrial production are higher.Traditional contact measurement has been gradually replaced by non-contact measurement because of its low work efficiency and possible damage to the surface of the object to be measured.Compared with other systems that can realize three-dimensional reconstruction,the binocular stereo vision system can complete the three-dimensional reconstruction of the spatial point without complicated light source equipment,and has the advantages of low cost and easy operation.This thesis mainly explores the problems of camera calibration,image correction,stereo matching and feature point 3D reconstruction in the binocular vision measurement technology of welded parts.Firstly,the coordinate system used in the visual system is introduced,and the construction of the experimental system is completed according to the principle of binocular stereo vision and the theoretical system.The accuracy of the calibration has a direct impact on the effect of 3D reconstruction.After discussing the commonly used calibration methods,the Zhang Zhengyou calibration method with high precision and easy operation is selected,and the distortion model in the calibration process is analyzed.Secondly,multiple sets of image pairs are collected by the binocular stereo vision system to be tested,and the image is processed by polar line correction.It mainly analyzes the features of the surface of the welded component to be tested in the processed image,and uses the SIFT feature matching algorithm to complete the matching work.Aiming at the problems of long matching time and incorrect matching,this thesis proposes improvements to the SIFT algorithm,and introduces RANSAC to eliminate mismatch points.Finally,in order to verify the effectiveness of the proposed algorithm,the three-dimensional coordinate reconstruction of the feature points of the welded parts on the image was successfully implemented on the software platform.Experiments show that the three-dimensional coordinates of feature points recovered by binocular stereo vision measurement technology have high accuracy.