Research On Circular Seam Tracking System Based On OpenCV

At present,there are many demands for large storage tanks in the fields of food processing and petrochemical industry,so it is difficult to automatically weld the surface annular weld in the tank forming process.Aiming at the problem of automatic welding of ring seam on the outer surface of wine brewing tank,this paper combines machine vision technology and welding robot technology.Through binocular vision system,the welding position is automatically identified,and then the welding robot is controlled to track the welding,so as to achieve the goal of improving the quality and efficiency of ring seam welding.First of all,according to the characteristics of the larger size of the brewing tank and the working principle of the vision sensor,the overall scheme of the system was determined,and the hardware platform of the annular weld tracking system was designed and built.On this basis,the camera calibration algorithm is studied to build the mathematical model of the two-dimensional coordinates of the image and the three-dimensional coordinates of the object.By improving the Zhang Zhengyou calibration method,the dot lattice calibration method based on Blob detection is proposed.The calibration experiment results show that the calibration accuracy of this method is high.At the same time,stereo correction is completed based on Bouguet algorithm,which realizes line alignment of left and right images,and corrects the binocular camera into an ideal parallel binocular system.Secondly,in order to solve the impact of uneven illumination on weld image,a variety of image enhancement algorithms were summarized and analyzed.MSRCR algorithm based on Retinex theory was selected to improve the brightness and contrast of weld image and reduce the difficulty of weld feature extraction.The weld area needs to be located before feature extraction.This paper proposes a method based on Hough line detection to locate the weld area and eliminate useless background.Aiming at the shortcomings of the traditional skeleton extraction of weld line,a distance transform method was proposed to repair the breakpoint and extract the single pixel weld,so as to obtain the pixel coordinate of the target weld.Then,the dense feature matching algorithm and the sparse feature matching algorithm are analyzed and summarized.Based on the SIFT algorithm and the geometric constraints of the opposite pole,the three-dimensional coordinates of the welding seam are obtained.Then,through motion analysis,the weld coordinates are converted to the welding torch coordinate system,and the tracking control strategy is designed based on the open architecture of PC+ motion controller.Finally,this paper uses OpenCV and Python to write the visual measurement system software.Combined with the motion control software written by Lab VIEW,the precision experiment and welding experiment of the annular weld tracking system are carried out to verify the feasibility of the tracking system.