Image segmentation via fuzzy additive hybrid networks

We have recently introduced a class of additive hybrid fuzzy set theoretic operators which can be inserted into neural network structures and trained to learn multicriteria decision making functions through a backpropagation-type algorithm. In this paper, we examine the use of such networks to aggregate evidence and perform supervised image segmentation. Such algorithms not only group like pixels together, but also provide class labels due to the training. The resultant network is used to segment a color image of a natural scene