AN AUTOMATED SYSTEM FOR THE RADIOGRAPHIC INSPECTION OF CASTINGS

Abstract This paper describes a novel approach to the design of an automated system for conventional radiographic inspection of cast engineering components. The system incorporates a series of conventional image processing techniques to efficiently detect and analyse defects in the radiographic image. These are combined with an intelligent knowledge based (1KB) system, which uses a series of rules defined by expert inspectors, to interpret and evaluate defects against the quality assurance of the inspection process. A description of the overall design of the system is given, including the various image processing techniques used. The elements of the 1KB system are described, including the knowledge representation scheme, inference mechanism and control structure, and rules for classification and evaluation. Preliminary results are given indicating that the system is capable of meeting the requirements of an automated system for industrial radiographic inspection. KEY WORDS: Automated radiographic inspectionimage processingintelligent knowledge based systems Additional informationNotes on contributorsA. KEHOE Address correspondence to: Dr. A. Kehoe, Department of Mechanical Engineering, University of Surrey, Guildford, Surrey GU2 5XH