The Research Of Disk Surface Defect Detection Technology

The research object of this thesis is the Nd Fe B material disk. if the disk has chipped edge, pitting or scratch defects on the surface, it will lightly influence on business value, or even cause serious consequences in use. In the current quality inspection also rely mainly on manual testing, but as a result of artificial detection of people mistakenly identified, due to subjective factors of targets, and the artificial detection efficiency is low, the labor cost is high, it is not conducive to the long-term development of the enterprise. Therefore using the image processing technology to defect the disk surface detection, which has important practical significance and economic benefits.In this thesis, disk surface defects to be detected have chipped edge, pitting and scratches. the method detecting the chipped edge of the disk is it, Firstly single threshold method is carried out on the disk image to generate the binary image, then template positioning algorithm is carried out on the binary disk image for locating, let the root coordinate system of image transforms to the fixed coordinate system in order to test the disk under the unified coordinate system; Then using the maximum entropy method extracts the target area to test for a disk, and using straight line fitting method based on least square fits the linear edge and circle fitting method based on least squares fits the rounded edge, calculating for each boundary point to the distances of the border; Finally, let the standard value compare the calculated distance to the judge whether the disk has chipped edge. The method detecting the pitting and scratches flawed of the disk is it, Firstly using template locates the position of image and converting it to fixed positioning coordinates, then disk image converted by median filtering process to remove salt and pepper noise interference image; Analyzing the pitting and scratches features of the disk image to set the appropriate gray threshold, and then determine the threshold for the image defect area is divided magnet pieces if the calculated defect or defects length; Finally, comparing with the value of the standard to determine whether it has pitting or scratches.Finally, I design and implement a disk surface detection detected system based on the above and test it. The results of implementation show that the detection rate of the surface defect detection system disk knock edge defects can reach 90%, for the detection of pitting or scratches defective disk surface is 71%.