A co-operative application of expert systems for photoelastic fringe pattern analysis

Abstract

This paper provides an account of a research project into the automatic analysis of the fringe patterns from photoelastic stress analysis. The work aimed to avoid expensive high-resolution image-processing hardware and retain compatibility with existing manual polariscopes. The paper describes how low-level (algorithmic) image processing is used to provide initial results. These results are both enhanced and analyzed by the use of knowledge-based techniques, resulting in a complete fringe pattern, incorporating fringe order information. A methodology was employed to systematically combine the elements of image processing, knowledge-based systems and user perception. Two resulting systems are described, together with the results obtained.