A fuzzy approximate reasoning model for a rule-based system in laser threat recognition

Intrinsically, vague knowledge is typical of the decision process especially when it is based on complex and non-numeric data. Linguistic descriptions play a significant role in decision making for pattern recognition. Fuzzy set theory and fuzzy logic provide an approximate framework for the representation of imprecise concepts and imprecise models of reasoning. In this article, fuzzy logic is proposed as a means of handling uncertainty in an expert system structure for automatic laser threat recognition which is increasingly required in modern war. Fuzzy characteristics of the feature values of laser threat patterns are comprehensively described. A prototype system has been developed and tested on labeled sample data of laser threats. The results are compared to our previous work, a similar recognition system. (C) 1998 Elsevier Science B.V. All rights reserved.