Research On The Method To Locate Welding Seams Based On Binocular Stereo Vision

As one of the most important technologies in modern manufacturing industry,welding technology is more and more widely used in aerospace,national defense,chemical industry,electronics and other fields.From the early manual welding to the present application of welding robots,welding technology has been greatly improved in both efficiency and accuracy.With the increasing demand for welding,welding robots are bound to become more intelligent and autonomic.Binocular Stereo vision,as a common and excellent visual sensor,has been gradually applied to the localization and navigation of welding robots.Based on the background above,this paper determines that the subject is Research on the Method to Locate Welding Seams Based on Binocular Stereo Vision.The research's focus is to recognize welding seams' location on the satellite plate by binocular stereo vision system.According to the background and content of the research,after analyzing the subject's demands and system's technical indicators,the whole scheme of binocular stereo vision system is determined,which includes the system hardware platform and software algorithm composition.The hardware platform mainly includes binocular cameras,a digital image acquisition card,an experiment platform and a computer.The software algorithm is mainly composed of camera calibration,stereo correction,image reprocessing,feature extraction and matching,binocular distance measurement,template matching and so on.Fully studying the camera imaging models,experiments of camera calibration are completed.Then the internal and external parameters are used to complete the experiments of stereo correction.Aiming at the characteristics of the titanium alloy pipe and its welding seam,image preprocessing of ROI region selection and image filtering is carried out.All the contents above are the foundation of binocular stereo vision system,which prepare well for the latter jobs.On the premise that the plane of the titanium alloy pipeline is parallel to the one of binocular vision system,this paper divides welding positioning into two steps.First,recognize the distance between the pipeline plane and the vision system plane,which is also called the identification of depth information.Then,recognize the coordinate information of the welding seam.Combining the depth information with the coordinate information,the position information of welding seams can be obtained.This topic studies stereo matching of binocular stereo vision.First,the topic compares SIFT and SURF of feature matching and chooses SIFT with a better result to extract and match features of titanium alloy pipes in the image pairs in order to get disparity information.Then the distance information between the pipeline plane and the vision system plane is calculated according to the principle of binocular distance measurement.In this project,titanium alloy pipes with different sizes and actual depth are tested.The experiments show the accuracy of measurement error can reach within 2mm,which meets the demand of technical indicators.Finally,this topic completes the coordinate information recognition of welding seams.First,the location of titanium alloy pipe is extracted based on the method of projection.Then the principle and methods of template matching are introduced.Finally,a two-step positioning method is used in the experiments.Firstly,the coordinate region is matched by the built template library.Secondly,the corner detection is detected by Shi-Tomasi algorithm as the coordinate information of welding seams.The experimental results show that the errors of horizontal and vertical coordinates are within 3 pixels,that is to say,the accuracy of measurement error can reach within 1mm,which meet the experimental requirements.