Research Of Measurement Technology For Shaft Parts Based On Convergent Binocular Vision

Shaft is an important part widely used in modern mechanical manufacturing industry.With the rapid development of intelligent manufacturing and automatic assembly technology,higher requirements are put forward for the accuracy and automation of measurement technology of shaft parts.The traditional contact measurement methods are inefficient,unable to carry out on-line real-time detection,and easy to scratch the surface of parts,which limits their application greatly.Since the 1970s,basis on the continuous improvement of electronic,optical and computational theory,the visual measurement technology has gradually improved and been widely used.Visual measurement method makes use of the self-scanning,high resolution and high sensitivity of the camera itself.It does not need to configure complex mechanical motion structure,reduces the source of error,and has the advantages of high accuracy,high speed and easy integration.In view of the above situation,this paper proposes a measurement method of axle parts based on convergent binocular vision.The specific research contents include:The first chapter summarizes the research status of key technologies such as non-contact measurement technology,machine vision based measurement technology and shaft parts measurement method at home and abroad,expounds the research background and significance of this paper,and introduces the research content and organization structure of this paper.The second chapter introduces the coordinate systems commonly used in the vision system and the transformation relationship between them.It expounds the basic principles of parallel binocular vision system and convergent binocular vision system.Based on the comparison of the advantages and disadvantages of the two systems,the vision system of this paper is built.An improved corner extraction method of calibration plate based on self-defined template convolution and energy growth rule is proposed.On this basis,a convergent binocular vision system calibration method based on 3D point cloud error is proposed.In the third chapter,aiming at the task of target detection of axle parts,a method of axle profile detection based on adaptive threshold and Hu invariant moment and a high precision method of axle ellipse end surface detection based on arc selection and combination strategy are proposed.The method can detect the contour of axle parts efficiently and lay a foundation for further geometric dimension measurement and attitude estimation.In the fourth chapter,the method of reconstructing axle profile based on spatial line reconstruction technology and the method of reconstructing axle end surface based on geometric measurement model of circular feature are described.On the basis of analyzing and comparing the advantages and disadvantages of the two methods,an automatic measurement scheme is proposed,and a complete measurement experiment is carried out to verify the validity and accuracy of the method.In the fifth chapter,combined with the methods described in this paper,a measurement system of axle parts based on convergent binocular vision is developed by using C++,OpenCV,Qt and other tools.The system includes camera calibration module,contour detection module and parameter measurement module.The system is used to measure many axle parts effectively.The sixth chapter summarizes the research results and conclusions of this paper on camera calibration in binocular vision measurement of axle parts,analyses the shortcomings of this paper,and looks forward to the future research directions.