Interval type-2 fuzzy logic for image edge detection quality evaluation

Recently there have been different methods to evaluate the edge detection of an image; most of them measure the similarity with respect to a reference image. In this paper we show the design and tests of a new method for edge detectors evaluation consisting of an interval type-2 fuzzy inference system, which inputs correspond to a combination of parameters that represent the most influential characteristics of an edge image according on previous experience. Each of these selected parameters was selected from existing methods for evaluation of edge detection. This new method is able to evaluate any edge detection process, including the traditional and fuzzy methods, but it was applied in synthetic images because of the need of an edge reference image for the input parameters calculation.