Research On Seam Tracking Based On Laser Vision Sensing

Welding is widely used as an important material forming and processing technology in the industrial field.The quality of welding directly determines the reliability and safety of welding engineering.The traditional manual welding method has become a welding development direction due to the harsh working environment,low production efficiency and high technical requirements for workers.In automatic welding,the weldment is affected by thermal deformation,which tends to cause the welding torch path to deviate from the actual weld trajectory,so that the welding quality is not guaranteed,and even the weldment is scrapped.Therefore,it is necessary to identify and track the weld position in real time and accurately.Laser vision sensing technology is widely used in the field of weld seam identification and tracking due to its simple structure,strong anti-noise ability and high measurement accuracy.During the welding process,due to the influence of arc,splash and dust dust,it is easy to cause the information of the image weld seam collected by the visual sensor to be seriously submerged,which affects the extraction of the weld feature points.Therefore,how to improve the sensing system and post-processing applications to accurately identify the center position of the weld is an urgent problem that needs to be solved urgently.This paper studies the welding feature extraction and tracking method and tracking system based on laser vision sensing.This topic first establishes a set of weld seam tracking test system,including the overall design of laser vision weld seam tracking system,the optical design of laser vision sensor and the selection of system related devices,and then establish the corresponding mathematical transformation model according to the actual system construction.And calibration of the internal and external parameters of the sensor to obtain the required parameters,to achieve the conversion between the pixel coordinates of the feature points in the weld image and its world coordinates.The current image processing technology is analyzed and studied.The weld feature point detection method based on Shi-Tomasi corner detection and the weld feature point extraction method based on maximum distance search algorithm are studied.In order to effectively realize the extraction of weld feature points,the image pre-processing operations mainly include image ROI(ROI: Region of Interest)extraction,image filtering,image segmentation,area threshold filtering,morphological trimming and centerline extraction.In order to improve the influence of the pixel points with large gray value on the calculation results of the weld stripe center in the stripe region,a boundary-constrained gray-scale weighted center of gravity method based on the extreme value method is proposed to extract the laser stripe center.For reduce the influence of various noises on the feature information extraction of Vshaped welds,a system state optimal estimation based on Kalman filter is used to predict the center position state deviation of the welds in order to reduce the influence of various noises on the extraction of weld feature information.The weld seam position information is obtained by processing the weld image,and the object state equation and observation equation of the system are established on the assumption that it is a state vector.The Kalman filter algorithm is constructed to obtain the optimal estimation of weld position deviation under the minimum mean square error(MMSE).The experimental results show that the Kalman filtering algorithm can reduce the influence of noise on the tracking system and improve the stability and accuracy of the tracking system.For straight-line lap-welded seam,after image pre-processing,the method of combining Shi-Tomasi corner detection with LK optical flow target tracking is used to realize the accurate recognition of weld seam feature points.