Research On Machine Vision-based Precision Testing Method For Mechanical Parts

With the continuous innovation and upgrading of modern industrial product design and manufacturing technology,the machinery and equipment manufacturing industry has experienced unprecedented rapid development.The detection of mechanical parts on industrial processing sites has gradually shown that the detection environment is complex,the detection objects are diverse,and the detection process tends to be automated and the integrated characteristics of the detection function.So that the traditional conventional size detection system and method have been unable to meet the increasing detection requirements of the machining accuracy of mechanical parts on the industrial processing site.In order to solve the problems of high precision detection requirements and low efficiency of mechanical parts detection,this article mainly completed the following work:(1)We explain the research background and significance of this article;introduce the composition and working principle of the visual inspection system;generally summarize the development status of visual inspection technology of mechanical parts at home and abroad;forecast the development trend of visual inspection technology;research the various image processing methods of visual inspection technology,and conduct relevant experimental analysis.Certainly,the image processing methods required in this article were determined according to the experimental comparison results.(2)We design and build a visual inspection platform according to the needs of experimental research.The main functions and components of the visual inspection platform are summarized,the selection of each part of the device is introduced in detail,and the model of the device used in the test is determined.(3)The dual-target calibration principle of industrial cameras is studied,and the parameters of the industrial cameras used are calibrated,which reduces the influence of industrial camera distortion on the test accuracy during the test in this paper.(4)We propose a method of mechanical parts processing accuracy detection based on machine vision technology for dynamic registration of images and graphics.The theoretical principle of the processing accuracy detection method is elaborated in detail.And combined with the improved SIFT-Harris image mosaic algorithm,the feasibility of the detection method is tested and verified.Through a reasonable analysis of the obtained test data,it is determined that the test accuracy of the method in this paper conforms to the criterion of normal 3? distribution error.