Detection and Classification of Surface Defects of Cold Rolling Mill Steel Using Morphology and Neural Network

As manufacturing speed increases in the steel industry, fast and exact product inspection becomes more important. This paper deals with defect detection and classification algorithm for high-speed steel bar in coil. We enhance an acquired image by use of a special subtractive method and find the position of defect using local entropy and morphology. The extracted statistical features are then presented to a classifier. We use neural network and fuzzy inference system as a classifier and compare their results. The best accuracy, %97.19, is obtained by the neural network.