| The increasing development of modern welding automation level and the advancement of computer vision technology have put forward high requirements for the quality of automatic welding welds.The accurate identification of the weld seam is an important guarantee for the quality of the weld seam.The weld seam identification technology is the core of the realization of automatic welding,and its research is of great significance.Therefore,this paper designs and builds a welding seam identification measurement system based on the line structured light sensor,which gives the welding robot the visual ability,so that it can accurately identify the location of the welding seam and complete the welding seam identification task.This paper mainly focuses on key technologies such as calibration of measurement system,centerline extraction of linear structured light stripes,and feature point extraction of weld images.The main content of this paper can be divided into:Firstly,the calibration method of line structured light measurement system is studied.For the calibration of camera internal parameters and linear structured light plane,a calibration method based on blanking points is proposed,which can obtain camera internal parameters and light plane parameters at the same time.The calibration method uses two fixed points in the space to control the robot to do two linear independent translation movements,to calibrate the rotation matrix,and four arbitrary pose movements to calibrate the translation matrix.The calibration experiment verifies that the method has high calibration accuracy and simplifies the calibration process.Secondly,the centerline extraction algorithm of linear structured light stripes is studied.In order to solve the problem of low accuracy of centerline extraction for a single weld type by existing algorithms,an improved grayscale centerline extraction algorithm was proposed.Firstly,the center line of the structured light stripe is initially extracted by the geometric center method,and then the normal direction of the initially extracted center line is judged based on the direction template method.,and finally obtain the centerline of the linear structured light stripe.The experimental results verify that the algorithm can accurately extract the centerlines of the four types of welds.Then,the method of feature extraction of weld image and identification of starting point of weld is studied.In terms of feature extraction of weld images,the least squares straight line fitting method and Shi-Tomasi corner detection method are used to complete the feature extraction of continuous and discontinuous welds.Identify the method to complete the starting point extraction.The experimental results verify that the proposed method can accurately extract the features and starting points of continuous and discontinuous welds.Finally,a welding seam recognition system based on active vision is constructed,which realizes the functions of image acquisition,processing,system calibration and welding seam feature recognition.The system was used to identify and measure the T-shaped and butt welds.The average error of the weld gap identification was less than0.15 mm,and the average error of the feature point extraction accuracy was less than0.11 mm.The experimental results show that the welding seam identification method can quickly and accurately identify the welding seam. |
