Reserch On Seam Recognition And Tracking Of Stainless Steel Box Based On Machine Vision

Welding is an important process in the manufacturing process of stainless steel box,which is generally completed by manual or teaching welding machine.Due to the influence of individuals,the manual welding method has the disadvantages of unstable welding quality and low welding efficiency;during the teaching welding,the shape of the box body changes slightly due to the clamping of the box body and the effect of external force,which requires repeated teaching and inconvenient use.In order to improve the welding quality and efficiency of stainless steel box,a three-axis welding experiment platform is built in this paper.The machine vision method is used to identify and process the weld,and the automatic positioning and welding functions are realized.Firstly,the mathematical model of the camera imaging system is established,and the calibration methods of the camera's internal and external parameters are studied.By using the camera with fixed attitude to take the image of the welding target,the corresponding relationship between the collected image and the actual welding point coordinate is obtained.Secondly,aiming at the defect of losing edge details easily in Canny's algorithm,the bilateral filtering algorithm is introduced to effectively retain the image edge information.In order to solve the problem of image discontinuity,edge connection,edge expansion and edge thinning are used to get a continuous single pixel edge.The simulation results show that the improved Canny algorithm can retain more edge details and get more continuous welding image,which is conducive to the location and control of subsequent welding points.Finally,according to the results of weld identification,the positioning control and automatic welding of stainless steel box are realized.At the same time,it analyzes and classifies different types of weld defects,evaluates the welding quality,summarizes the imaging characteristics of typical weld defects in the image,realizes the real-time marking of weld defects,reduces the cost of manual detection,and improves the work efficiency.