期刊名称：International Journal of Hybrid Information Technology
印刷版ISSN：1738-9968
出版年度：2016
卷号：9
期号：1
页码：253-266
DOI：10.14257/ijhit.2016.9.1.22
出版社：SERSC
摘要：The research of defect detection technology of magnetic sheet surface has important significance for improvement of production efficiency and product quality in enterprise. A single threshold method is used for binary segmentation of magnetic sheet image in this paper, then the template localization algorithm is used for template position of magnetic sheet image after binary segmentation, the magnetic sheet image with different position should be transformed into a unified coordinate system for detecting, then least square method is used for fitting straight line and fillet edge, then the calculated values are compared with standard value to determine whether are knock edges. Further, the appropriate grayscale threshold is set by the pitting and scratches characteristics of magnetic sheet image, the magnetic sheet image should be segmented according to the determined threshold value, if the defect exists, the defect area or length will be calculated, then the result is compared with standard values to determine whether is pitting or scratches. The final tests of technology in this paper show that the detection rate of knock edge and the detection rate of pitting or scratches are both acceptable.
