Research On Automatic Detection System Of Inner Surface Defects Of Sliding Bearing Based On Machine Vision

For the factory inspection of the sliding bearing coated with the wear-resistant layer,the main test item is the surface defect of the wear-resistant coating as the bearing working surface.At present,there are many disadvantages of manual testing,which can not meet the demand of productivity and customer.Based on this,this thesis put forward a kind of overall engineering solution for the automatic detection of the sliding bearing surface defects based on machine vision.Based on the industrial application of machine vision technology,the solution of optical electromechanical integration system was designed.Based on the characteristics and testing requirements of the inspection project,the detection algorithm of inner surface defect was designed and verified experimentally.By analyzing the characteristics of the defect image,a defect classification algorithm based on SVM classifier was designed and verified experimentally.Through the integrated development of the software and the cooperation of the automation technology,the purpose of detecting the defects of the inner surface of the sliding bearing was realized.Specific research contents are as follows:(1)In this reseach the optical electromechanical system was designed.Through the analysis of the production environment and process of the sliding bearing,according to the detecting requirements and the automation requirements,the overall design scheme of the detection system was put forward.According to the accuracy requirements and product cost,the software and hardware involved in the scheme were selected and designed.Build the experimental detection platform,to achieve the high-definition image acquisition of the inner surface.of sliding bearing.(2)In this reseach the algorithm for detecting the defects of the inner surface of sliding bearing was designed and verified experimentally.Firstly,the general scheme of defect feature extraction algorithm was developed by analyzing the characteristics of defect morphology.Secondly,the inner surface image was calibrated by the calibration plate,and the defect size was quantized.Thirdly,the segmentation of the inner surface ROI was realized based on the template matching algorithm and the threshold segmentation method based on the global gray value.Fourthly,the edge defects were extracted by using the image processing method based on region morphology and image subtraction;the scratch defects were extracted by using the image processing method based on Gabor filter and the improved image gradient and Otsu algorithm.Finally,the accuracy and feasibility of the algorithm were proved by the experiment.(3)In this reseach the defect classification algorithm was designed.By analyzing the geometrical features of the defect,the shape feature and the texture feature totaling the 12-dimensional eigenvector,a set of the characteristic parameters of the inner surface defect of the sliding bearing was established.According to the classification experiment,the appropriate kernel function and multi-classification algorithm was chosen to construct SVM classifier.Through the comparison test of two multi-classification algorithms,the OVR-SVMs classifier was selected from the aspects of detection accuracy and time-consuming.The detection accuracy can meet the actual classification requirements.(4)In this reseach the system integration and experimental analysis was completed.Based on the C++ programming language and the Halcon image processing library,the system software was integrated and developed through the modular programming idea,and a complete detection system was built.In the experimental stage,the accuracy rate of the machine detection was 98.39%,which was 4.04% higher than that of the artificial detection,and the average detection time was within 700 ms.The superiority of the machine detection was demonstrated.This thesis provided a useful exploration for solving the automatic detection of the working surface defects of the sliding bearing coating,and put forward the ideas and concrete solutions for the engineering application of the machine vision inspection.