Study And Development On Size And Surface Defect Inspection System Of Magnetic Ring Based On Machine Vision

In order to meet the requirements of high efficiency,high precision and intelligence,as one of the non-contact detection methods,online detection technology based on digital image processing technology is more and more used in various fields of industrial production.Based on this background,this thesis studied the Ring size measurement and surface defect detection method.A series of dimension and surface defect detection algorithm based on machine vision is proposed and applied to the on-line detection system.According to the inspection requirements and production environment,this thesis designs an on-line defective magnet ring detection system.The mechanical part,visual part,control part and the application software are included in this system.The Mechanical parts to complete the feeding,conveying,flip surface tasks and so on.The visual part completes the original picture acquisition.The control mechanism receives the sorting signal to eliminate defective ring,etc.The application software is responsible for processing and identifying the acquired images and sending the sorting instruction.Through the analysis of the characteristics of magnetic ring image,Spatial filtering,morphological processing and threshold segmentation are studied in this thesis.Through the comparative analysis of the commonly used spatial filtering methods,the improved median filter is used to filter the original image.Extracting the contour of the Magnetic Ring based on the binarization and morphological contour extraction algorithm.The minAreaRect function provided by OpenCV can be used to fit the side contour and the improved Hough ellipse-fitting is applied to fit the positive side contour.Then the size of Magnetic Ring is got by calibrating the Camera.Based on the ellipse and rectangle parameters information,region mask method is used to extract the foreground area of magnetic ring image.The foreground image is mapped from the sub peak gray level to the background gray level.Then we use a variety of adaptive threshold segmentation method to segment magnetic ring defects and do dome comparative analysis.Aiming at the case that the ring surface texture interfere the segmentation of defects,this thesis presents an edge segmentation algorithm based on wavelet analysis of texture weakening and adaptive Canny.Through the multi-scale wavelet transform,transform the image and extract the approximate coefficient of magnetic ring.In the study of the statistical feature based on gray level co-occurrence matrix and gray histogram,a certain function relation between the magnetic surface texture entropy and canny operator parameters was found.The threshold of canny operator is adaptively determined by this function relation.The edge of ring and defective region is detected by edge detection of approximate part in wavelet domain.Then morphological processing and post-processing of edge is done.The defective region edge is got by XOR between the edge achieved above and the contour.Judging closure and filling inside after closing the defective region edge,defect area is extracted.The results of different illumination show that the algorithm is robust to illumination.A large number of experiments show that the 94.7% ring defects could be detected by this algorithm.The application of magnetic ring surface defect detection software is developed on the basis of Sapera LT SDK and OpenCV computer vision library.Software development based on VS2013 environment and building an application UI based on MFC.Under the cooperation of PC software and the sorting pump,the system can accurately detect product intact or defective and automatic pick up the waste,achieving the goal of intelligent detection with magnetic ring.