Research On Weld Recognition Method Based On Structured Light Vision

With the rapid development of industrial automation technology,automatic welding has become an important part of modern welding process.The main advantages of automatic welding include improving production efficiency,reducing manual errors,reducing labor intensity and improving welding quality.The accurate identification of welds is an important prerequisite for high-quality welding,and is the key to realize automatic welding.This paper builds a weld recognition system based on structured light vision,combines machine vision with welding robot,and uses robot to drive the motion of the vision device to collect the images of the workpiece under structured light irradiation,and performs a series of image processing to finally complete the recognition of the weld and its feature points.The main research contents are as follows:(1)For the hardware part of the structured light vision system,this paper introduces the composition of the structured light vision system,completes the selection of key components of the structured light vision device,the structure of the fixture and the design of the weld identification scheme,providing hardware support for the task of weld identification.(2)For the calibration of structured light vision devices,an automatic calibration method based on non-special targets is proposed in this paper.By controlling the robot to drive the structured light vision device to carry out two sets of motion,the method takes pictures of the non-special target,processes the captured images,including image enhancement,feature point extraction and feature point matching,and finally uses the feature point information to complete the camera and light plane calibration.Experimental verification shows that the structured light device calibrated by this method has high accuracy.(3)For the identification of welds and feature points,this paper first proposed a center line extraction algorithm based on neighborhood polygon centroid fitting,and extracted the center line of the structured light fringe in the form of multi-segment straight line combination.On this basis,the welding seam identification method based on straight line composition was adopted to complete the identification of welds and feature points.The identification scheme of the initial feature points of weld is designed.The experimental results show that this method can extract the center lines and feature points of four kinds of welds accurately.(4)As for the software of the welding seam identification system,this paper developed the welding seam identification software based on structured light vision and used the software to complete the tasks of calibration of the structured light vision device,identification of weld types and feature points.The initial feature point verification experiment was carried out for four kinds of welds.Through the experimental verification,the average error of feature point extraction was less than 0.1mm,which verified the effectiveness of the proposed method.