Image Processing Based Defuzzification Method for Type-2 Fuzzy Systems

Abstract

Type-2 fuzzy logic systems are the systems that use the three dimensional fuzzy membership functions in order to minimize the effects of the indefiniteness not modelled by type-1 fuzzy systems. However, in type-2 fuzzy systems, the complexity in terms of calculation is much higher compared to type-1 fuzzy systems. Particularly the defuzzification stage of type-2 fuzzy systems is a complex calculation process. In this study, an image process based algorithm was suggested for the defuzzification process of type-2 fuzzy systems. The suggested algorithm is both faster and easier to realize compared to those used in the literature. The method is based on converting the area composed of the active rules into black-white image and finding the weighted mean of the black parts in the image. The effectiveness of the defuzzification method suggested was shown in the simulation results provided.