Research On Pipeline Welding Location Method Based On Binocular Vision

With the continuous development and innovation of science and technology,industrial robots have been widely used,and the robot industry is gradually mature.Since the emergence of industrial robots,the largest amount of use is the welding field.According to statistics,the operator in the welding process to determine whether the need to make adjustments on most of the information is derived from its own visual,machine vision and image processing technology in the welding field is equivalent to the operator's "eyes".The traditional welding technology is prone to thermal deformation,assembly error or artificial interference in the process of welding,which leads to the decrease of welding quality.Therefore,how to accurately locate the pipeline welding in the non-contact situation has become an important issue in the field of welding.Based on the visual target positioning robot welding line as the research background,the system calibration and feature extraction and template matching algorithm based on the theory of pipeline,workpiece on the binocular vision system to identify the location of the welding problem to do in-depth research and analysis,has certain theoretical significance and practical value.The main work of this paper is as follows:(1)Calibration of internal and external reference of camera and acquisition of world coordinates in binocular vision system.Through the analysis of principle and calibration of the camera imaging model of binocular vision,nonlinear used Zhang Zhengyou calibration method in HALCON programming environment,using the camera calibration of binocular vision system,get the camera internal and external reference.According to the internal and external reference of binocular camera and binocular vision principle,the depth information can be obtained from the intersection of the left and right two lines of sight,that is,the three-dimensional coordinate value in the world coordinate system.(2)According to the characteristics of the pipeline workpiece,image feature extraction and template matching.The industrial camera is used to collect the image of the pipeline workpiece,and the image is processed by weighted average and median filtering.Create a template using the HALCON programming environment,extract the edge of the template image and the reference point,and then use the HALCON method,the shape of the edge of the weld template in the image matching based on and reflect the matching effect in the image to be detected.The camera calibration results are substituted to obtain the depth information,and the three-dimensional coordinate values of the welding in the world coordinate system are obtained.(3)A binocular vision Hand-Eye calibration algorithm for three-axis manipulator is proposed and optimized.By analyzing two kinds of Hand-Eye calibration forms of Eye-in-Hand and Eye-to-Hand,the Eye-to-Hand form is selected according to the actual application environment.By deducing the principle of Eye-to-Hand form,the specific coordinate value of input data is changed into coordinate difference,and the iterative optimization is carried out.Calibration experiments using the optimized welding in Hand-Eye specific three-dimensional coordinate manipulator base coordinates obtained by camera calibration and template matching results,realize the positioning welding manipulator.