Donald Bren School of Information and Computer Sciences

This paper outlines the issues involved in automating nondestructive evaluation techniques. Nondestructive evaluation techniques are used to inspect a variety of parts during manufacturing and service. Currently, humans analyze the output obtained from test techniques by looking for features which indicate that a defect is located in the material. This evaluation is dependent on both the experience and alertness of the technician performing the test. Automation of these processes should improve the consistency of results and enhance the testing of more complex materials. Machine learning and pattern recognition techniques are being investigated to automate the process.