Research And Implementation Of Machine Vision-based On-site Inspection Technology For Mechanical Parts

With the advancement of science and production technology and the continuous upgrading of automation,machine vision is also playing an increasingly important role in inspection,replacing manual labor to ensure efficient,rapid and repeatable work,and has broad prospects in actual production and life.Based on the analysis of the research on the application of machine vision in modern production,this paper studies the in-position detection of mechanical parts on the production line.This paper takes gears and coated unfolding plates as the research objects,and carries out in-position inspections on them on the assembly line.Carried out the overall plan design of the in-position detection system,detailed the basic structure of the hardware system,including the camera,lens,photoelectric switch,My-Rio,cylinder and dark box,etc.,carried out the selection and construction of the hardware system,and designed the image acquisition and light source Illumination system scheme;propose an image preprocessing scheme.In image preprocessing,through image filtering,image segmentation,edge detection and other steps,a clear edge of the image is obtained.The imaging principle of the camera is introduced,the qualification test of gears is carried out,the detection method of the straight line automatic division of the coating unfolding plate is proposed,the automatic division detection is carried out according to the band shape and gray characteristics,and the combination of polynomial interpolation and least square regression fitting is proposed.To obtain the coordinates of each point,Labview and Matlab are combined to realize mixed programming,and the detected data is transmitted to the CNC machine tool through RS232 for real-time processing.Perform on-site experimental tests on gears and coated unfolded plates,compare the results of manual inspections with in-place inspection systems,and analyze the experimental results.Environmental errors,hardware errors,software errors and processing errors during the experiment Error analysis and optimization.The innovation of the in-situ detection technology for mechanical parts based on machine vision researched in this paper is the combination of Labview and Matlab for hybrid programming,and the design of an automatic partition method based on the combination of geometric shapes and gray features,and a polynomial design.The fitting method that combines interpolation and least squares regression,the machine tool obtains the detection data for real-time processing,can realize rapid identification and accurate calculation,and has high detection accuracy and detection speed.