Research On Surface Defect Recognition Method Of Magnet Device Based On Machine Vision

With the rapid increase in the output of magnet devices,traditional manual inspection can no longer fulfill the production needs.The emergence of machine vision has liberated people’s hands.It has the advantages of high accuracy,strong reliability,high safety and non-contact detection.At present,the domestic machine vision technology for detecting magnet defects has been developing,but there is still a lot of room for development in the industrial field,especially in complex factory environments,such as the dark background of the workpiece to be tested,and the relatively low contrast between the workpiece and the environmental background,etc.,The algorithm in this paper can also realize the detection function.In this paper,a set of magnet defect detection system is designed,with reasonable light source module and processing algorithm,it can detect four defect categories of scratches,corners,pits and rust spots in a dark environment,and realizes the classification and identification of defects and precise positioning.It has very good application value.In order to improve the contrast between the workpiece and the background,the images of different light intensity are merged to extract more defect details and effectively reduce the probability of omissions in defect detection.The low-angle illumination of the ring light source used in this paper can also avoid artifacts and shadows on the edge of the workpiece contour,and better identify the contour of the workpiece.The algorithm processing research is carried out on the surface defects of the magnet device.Adopt bilateral filtering denoising algorithm,which has the advantages of edge-preserving denoising;adopting image fusion algorithmhwhich can enhance the contrast of the picture and highlight the defect details;the gamma correction can adjust the detail information of the low-gray image;by adjusting the ACE algorithm Threshold value enhances the details of the picture;Canny edge detection and morphological closing operation make the defect contours well closed.The contour extraction algorithm is designed,including the contour finding algorithm and the contour drawing algorithm,and the contour filtering is carried out in the follow-up to remove the multi-layer contour problem and some small environmental interference.By calculating various characteristic parameters,different types of defects are distinguished according to different aspect ratios,filling degrees and gray averages.Finally,a defect location algorithm is designed,which can realize the precise location of the defect,and the defect category is also marked near the image defect.The location and category information of the defect can be seen intuitively.Finally,an experimental platform was built to verify the algorithm of this paper,and a host computer software based on LabVIEW was designed to realize human-computer interaction,which can display the classification information and characteristic parameters of defects on the interface.Through experimental verification,the accuracy of this system can reach 0.1mm.