Knowledge-based hybrid expert system for automated interferometric data reduction

Abstract

In order to accomplish the ultimate goal of automated interferogram evaluation, the utilization of artificial neural networks or expert systems in artificial intelligence has been expected. In this paper, a new PC-based fringe evaluation technique, which is based on hybridization of an expert system and a conventional fringe tracking method, is presented together with some test results of noise-ridden interferograms in high-speed aerodynamics. This technique incorporates algorithmic image processing with expert-system-driven knowledge manipulation in a single structure to provide an optimal solution for noisy interferogram evaluation. A new approach for regional phase unwrapping based on the technique is also proposed. This concept is applicable to interferometry such as phase shifting and Fourier transform that provides phase-wrapped information. It is based on conservative decision making in defining phase-wrapped regions by utilizing global/regional knowledge in the expert system. The approach is thus believed to reduce computation time and memory space in processing complex or noisy interferograms that require frequent operator interactions.