Laser Scanning Weld Image Processing And 3D Reconstruction

In the industrial circle,Welding technology is an very vital component.Postwelding inspection technology is usually used to check whether the quality of welding is good,so post-welding inspection technology has been improving with the progress of welding technology.Early post-welding detection was directly observed by human eyes,with low intelligence,and now the welding environment is more and more complex.Some relatively harsh and complex welding environments can affect the inspection efficiency and health of inspectors.Since all the information of postwelding detection is collected by visual principle,and now the development of machine vision technology is more and more advanced,Machine vision technology can be used for post-welding inspection can better deal with various welding environments instead of human eyes.After research on domestic and foreign post-weld inspection technology,this paper designs a set of laser scanning welding seam shape detection system for cylindrical welding parts with narrow welding environment,concave welding seam and difficult to be observed by human eyes.System consisting of upper machine and lower machine,a machine under direct optical triangulation method is used to obtain the depth of the sample point weld feature information,PC to realize the process of human-computer interaction and image processing,the next bit machine with mechanical arm welding torch welding at the same time,on the surface of the butt welding of interval continuous shooting,such as after completion of welding,PC receives the collected images,And complete the pretreatment work and the Central Line extraction,feature extraction of the extracted Central Line,the obtained feature points for coordinate transformation,and finally complete three-dimensional reconstruction.The focus of this system is on the image processing and 3D reconstruction of laser scanning welds.In the image pre-processing stage,there are mainly three steps.First,the binarization processing of weld images and the extraction of ROI regions are carried out to reduce the amount of calculation in image processing.A denoising method combining gaussian filter and median filter is determined.The center line of laser fringe is usually obtained by Steger algorithm,Harris corner detection was used to detect the transverse edge of the center line,the slope analysis method was used to extract the features of the detected middle line segment,and the 3d coordinates of the sampling points were converted by the parameters calibrated by the system.QT was used to call open source libraries such as Open CV and Open GL to achieve the reconstruction effect.The laser scanning welding seam shape detection system in this design meets the actual needs in both automation and safety,and can cope with a variety of welding scenarios,with good applicability and expansion.