Automated surface inspection system for black resin coated steel

This paper presents defect defection and classification methods for back resin coated coil in the steel industry. The detection algorithm is based on second order statistics of images. To discriminate the detected objects into defect classes, we use support vector machine (SVM). The total 20 attributes are extracted from each defect. To select best model for SVM classifier, we search the parameter spaces by grid search method. The experimental results show that the detection rate is over 98% and classification rate is over 90%.