Fuzzy pattern matching with adaptive bin width histograms to maximize classification performances

Fuzzy pattern matching is a supervised classification method, which uses histograms to generate membership functions. The classification performances increase with the separability between classes. We propose to use histograms with adaptive bin width according to the dispersion of learning samples in each class. The goal is to increase the separability between classes and to reduce the classification time. The efficacy of this method is tested in using several real examples.