Research And Development On The Defecting Surface Detections Of Magnetic Ring Based On Machine Vision

The research object of this paper is the magnetic ring used in the magnetron of the microwave oven.At present,the detection of magnetic ring surface defects mainly relies on eye detection,which has low detection efficiency,low reliability and high labor cost.Machine vision is a method of locating,recognizing,detecting and measuring objects by machine instead of human eyes.The method of machine vision can be used to realize automatic detection,improve production efficiency and product quality.In this paper,a machine vision based magnetic ring surface defect detection system is designed to improve the efficiency and reliability of magnetic ring detection.This paper mainly carried out the following work:(1)Defect analysis was carried out for the magnetic ring of the detected object,and the overall hardware scheme was designed based on the detected object.The design method of 8 stations was adopted to detect the chamfered surface,nonchamfered surface,inner completeness and outer circular surface of the magnetic ring.According to the experimental results,parallel plane light source,annular light source and dome light source were selected in different stations,and the visual selection scheme of each station was designed.(2)In this paper,the preprocessing methods used in image processing are studied.The original image of magnetic ring is preprocessed by image enhancement,filtering,image segmentation and image morphology processing,which improves the image quality and facilitates the extraction of magnetic ring defects.(3)Defects were extracted according to the defect of the circular features,gray feature,texture feature and shape feature,adopt the method of principal component analysis to choose contribution to defect features higher characteristic vector,using support vector machine(SVM)method for the circular defect classification,finally achieved the impellers,impurities,crack,and mark the classification of the four types of defects,the classification accuracy of 95% or more.(4)The software part of the system is designed,which includes three modules: camera image acquisition,image processing and defect recognition,and human-computer interaction interface.Finally,the whole process of the magnetic ring surface defect detection system is realized.The effect of the system on the magnetic ring surface detection has reached the real-time requirements and detection accuracy required by the manufacturer,and the accuracy of defect classification has reached more than 95%,saving labor and improving production efficiency.