Surface Defects Detection of Strip Steel Based on Semi-definite Programming SVM

摘要

The point at issue is to cure the problem of low calculation speed, lower recognition accuracy in the surface defects detection algorithm for strip steel. A novel surface defects detection method of strip steel based on Semi-definite Programming SVM (SDP-SVM) algorithm which is going to be proposed. The whole process of the defects detection algorithm includes extracting stripping steel image, ascertaining of defect image, preprocessing defect image, extracting defect features, classifying defects by SDP-SVM and confirming defect types, etc. Finally, simulate various defect images in the process of actual production which is based on the defects detection algorithm proposed in this manuscript. The results show that the algorithm has greatly improved compared to the traditional SVM algorithm both in accuracy and speed. The algorithm has a widespread application prospects with a more comprehensive approach, precision in application and practical utilization.