Determination of threshold values of membership functions is one of the key stages in design of a fuzzy logic inference system. This paper presents a statistical-based method using probability density functions to simplify the construction of membership functions and to fine-tune the critical points of the membership functions for the input variables. A Mamdani-type fuzzy inference system was developed to classify potato crop into healthy and damaged groups based on the image texture features. The results have shown a promising performance of the proposed method for design of fuzzy logic classifiers.
