Research On Recognition And Positioning Of Small Casing Weld Based On Machine Vision

With the rapid development of industrial automation welding technology in country,a large number of welding robots have been widely used in welding various types of welds.At present,the existing technology can identify the weld well which is obvious and have Fixed position,however,the automatic welding technology of the small casing weld with the complex background is still not achieve,one of the main reasons is that the identification and positioning of the small casing weld is difficult to realize.This article focuses on the welding problem of the small casing welding seam,and conducts research on image processing technology and visual measurement technology.The main research contents are as follows:First,this thesis introduces and summarizes the development status of automated welding systems,weld recognition technology,and three-dimensional measurement technology.On this basis,it analyzes the deficiencies of the existing technology in the detection of casing welds.Then,according to the complex environment of the weld,an image segmentation method is proposed based on the color features of the weld area,and the effects and characteristics of segmentation in different color spaces are compared.In addition,combined with the morphological operation and the basic operation method of the image on the binary image after image segmentation,the potential regions of the weld is determined,and according to the position and shape characteristics of the weld,the identification of the casing weld is realized on the image.In addition,in order to make the casing welding recognition method more robust,this thesis also conducted some research on the image preprocessing method.Before recognizing the weld,different image preprocessing methods are used for the images under different brightness conditions to improve the quality of the image to be recognized.Thirdly,the binocular vision measurement is used to positioning the welding seam according to the actual welding environment,which is calibrated by the Zhang Zhengyou calibration method.The binocular vision measurement model and commonly used image feature matching methods are introduced,the effects of different feature matching methods applied to casing weld feature matching are compared,and the epipolar constraint of simple feature point is used to match the weld.Finally,the binocular vision measurement model was used to positioning the weld,and the basic causes of the positioning error were analyzed.In this thesis,the visual recognition and positioning method of the small casing weld is studied,and the software for the recognition and positioning of this kind of weld is developed.Experiments show that the system can meet the visual positioning accuracy of the casing weld seam during the gas welding of the manipulator.