An architecture for a knowledge-based image inspection system

Defect classification by image based techniques is an important issue in quality assurance and nondestructive testing. The solution of the problem is usually complex and context dependent. A domain specific interpretation is required. For this not only knowledge about the appearance of the objects is necessary, but also knowledge about the technological background is required. Thus, the acquisition, representation and use of domain specific knowledge in combination with image processing facilities is a central point. The architecture of a knowledge based defect classification system is described. The architecture is flexible enough to be used for different applications. The performance of the system is described for defect recognition and diagnosis of misprints in offset printing