Pattern Recognition Based on Fuzzy Three-Valued Logic Reasoning
Cui-Yan, Jin-Yong Tao, Wang-Jun Tao · 2009
Fuzzy pattern recognition is an application of fuzzy mathematic theory in the field of pattern recognition. Theoretic basis of this method is fuzzy multiple-valued logic whose truth-value is determined depending on membership function. Now, however, the membership function definition of fuzzy set has not formed uniform method and general format yet. Determination of membership function is mainly affected by subjective factors and is easy to become troublesome problem. For solving this problem, this paper firstly proposes fuzzy three-valued logic [0, 0.5, 1] on the basis of research fruits about physiological function of "synapse" of human's cerebra and objective situation of truth-values of proposition existing. According to L. A. Zadeh's fuzzy set theory, synthesizing operation rules of fuzzy relationship and fuzzy reasoning theory, the paper introduces and discusses the explanation for fuzzy three-valued logic and the scientificalness and rationality for applying it into pattern recognition. Application examples in this paper indicate the feasibility and maneuverability of the method proposed in this paper. Certain practical value has been presented with the fact that the recognition results comply with the actual situation.