Research On Automatic Welding Detection System Based On Laser Binocular Vision

Welding as a low-cost,accurate and reliable method of connecting materials with high technology,has become an irreplaceable important processing technology in industrial manufacturing.To date,there are no other method than welding can be more widely used in the connection of metals,and add more value to the products welded.Since Germany proposed the concept of industry 4.0,the intelligent manufacturing of advanced industry has attracted more and more attention from more and more countries,and automatic welding technology has been applied more widely in manufacturing industry.In this paper,the automatic seam detection system based on laser binocular vision is studied to improve the flexibility and precision of automatic welding technology.Firstly,in order to collect clear laser stripe information of the weld,the overall structure of the welding automatic detection system based on laser binocular vision will be constructed,including the laser binocular vision system and the motion control system.Secondly,the monocular camera model and binocular vision system model will be theoretically analyzed,and the left and right cameras in the binocular camera model are respectively calibrated by Zhang Zhengyou calibration method and the calibration tool in matlab2015 b to obtain the internal and external parameters.Then the external parameters of laser binocular vision system will be obtained by calibrating the binocular vision system.In order to separate the laser stripe of weld seam from the background of weldment,the image preprocessing of the laser stripe is needed.Firstly,the color weld laser stripe image is converted into gray image by graying.Secondly,the pepper and salt noises in the laser stripe images are removed by the image filtering algorithm,and the median filtering algorithm is improved to shorten the running time of the algorithm.Then the image enhancement algorithm is used to enhance the laser stripe information and improve the contrast between the laser stripe and the weldment.Finally,the laser stripe image of weld seam is converted into binary image by threshold segmentation.By comparing the advantages and disadvantages of several laser stripe centerline extraction algorithms for weld seam,the direction template combined with ridge tracking algorithm is finally selected to extract the laser stripe centerline of weld seam.The sub-pixel corner detection algorithm is used to detect the pixel coordinates of the weld feature points.On this basis,weld feature points will be reconstructed in three dimensions to obtained spatial coordinates of the weld feature points,then the weld track is generated.