Visual defects classification system using co-occurrence histogram image

The Visual Inspection System is used on various production systems and that effectiveness is verified. The defects classification system for inspection system using neural network has been developed to improve that quality and proved its effectiveness. However, there are some classes of defects which are not detected with enough reliability using conventional systems. To solve the problem, the method using co-occurrence histogram image is proposed. Co-occurrence histogram can detect especially wide-spread defects. Our study focused to analysis co-occurrence histogram by image processing to obtain better recognition rate of defect classification. In this paper, the concept of the defects classification system using co-occurrence histogram image is described, and some experience has been introduced.