Research On Recognition Of Steel-Strip Surface Defects Based On SVM

With the flying development of our society, cold steel strips products is getting more and more widespread in the social product and the life application, so its quality becomes the focus that people pay close attention to day by day. However , the existing methods have many shortcomings, such as bad real-time performance ,low detected degree of confidence, formidable circumstance of detection, etc. So the research of steel strip surface defect become one of the focuses of technology researching at present.Surface defects recognition techniques based on SVM are further researched in this paper. The details are as follows:(1)Research on image preprocessing algorithm. This peaper used various image de-noising methods and evaluted its effects. Considering the experimental results and our practical needs to choose the median filtering methord.(2)Research on image edge detection algorithm. Using various image edge detection algorism and comparing the results to select the Canny operator algorithm.(3)Research on character extraction. We extracted various features ,such as hu moment invariants, geometric feature, gray feature, texture feature based on synthesized grey level co-ovvurrence matrix, texture featrue based on brightness histogram, we analysis on these features to choose the suitalbe feature as classification feature.(4)Research on classification and recognition algorithm. We used SVM and netural network methord to recognizing these defects. The experimental results showed that SVM algorithm has more advantages than netural network.(5)Research on classification and recognition algorithm based on fusion technology. Based on the analysis and test in the fourth chapter,using classification methord based on fusion technology. The experiment shows that it simplifies the calculation complexity effectively and makes system has complementarity.