City Research Online

Abstract

The thesis presents the results of a theoretical and experimental study on the application of syntactic pattern recognition to defect classification in optical surface inspection.

The inspected surface is interrogated with visible light and as a result of this interaction, the characteristics of the reflected light provide information about the surface. Signal processing techniques are applied to detect and delineate the defects on the surface. The syntactic pattern recognition system designed in the thesis operates on the specular videoprint, which is a black-white representation of the defects extracted from the specularly reflected light.

A pattern description language which is well suited for online processing is developed for the representation of two-dimensional defects as strings. The languages which describe the classes of defects are approximated by regular grammars obtained automatically using an efficient grammar inference algorithm. ‘The classification of the defects is achieved using an algorithm based on the concept of edit operations which reduces the errors due to the approximation of languages and to the imperfect structure of the defects. A set of programs written in FORTRAN is used for performing a detailed analysis of the system in computer simulation.

The validity of the theoretical methods proposed is tested on a data base of real defects, acquired using an automated data gathering system. ‘The experimental results obtained in simulation with this data base indicate an acceptable performance of about 70 - 80% correct recognitions. A hardware implementation of the classification system is outlined and shown to be economical and capable of operating in real time.

The interpretation of the results reveals the limitations of the present system and suggests techniques for further investigation to improve the recognition performance.

Publication Type:	Thesis (Doctoral)
Subjects:	Q Science > QA Mathematics > QA75 Electronic computers. Computer science Q Science > QA Mathematics > QA76 Computer software T Technology > TK Electrical engineering. Electronics Nuclear engineering
Departments:	School of Science & Technology > Department of Engineering School of Science & Technology > School of Science & Technology Doctoral Theses Doctoral Theses

Export

Downloads