Evaluation of edge detection parameters using fuzzy edginess

Most edge detection methods have parameters (threshold values or standard deviation of Gaussian operator for smoothing) to be set. These parameter values have much influence on the output of the detectors. In this paper, we propose a method of automatic parameter evaluation in edge detection. The parameters are evaluated based on the ambiguity of location and formation of edges. These ambiguity measures are defined by using fuzzy edginess. The most significant set of parameters can be obtained by selecting the values which give the minimum ambiguity. No iterative visual interaction or a priori information about the edges are needed. The effectiveness of this method is demonstrated through various experiments using synthetic and real images.