A new type of fuzzy rule-based system and its application to edge detection in images

A new type of fuzzy rule-based system is proposed and its application to image edge detection is presented. This technique utilizes human knowledge on how an input signal should be processed depending on several feature values about the input. Here, the knowledge is represented by fuzzy rules and at the same time represented by a multidimensional nonlinear function. This system automatically design the membership function including the rules by optimizing the multidimensional function for training data. Especially, the edge attracting system proposed here can be easily optimized without iteration. Computers simulations verify its high performance in edge detection