A new neural network model for image segmentation

Abstract

In this paper, the problem of image segmentation using artificial neural network ( ANN) is discussed. A new ANN model for image segmentation is proposed which consists of two sub-networks in cascade, i.e. the master sub-network (MSN) and the auxiliary sub-network (ASN). The dynamics of this model is studied in details and the corresponding algorithm is described in the paper. Finally, the simulations are carried out for the medical image using the moment-preserving thresholding and the proposed ANN model, and the corresponding results indicate that the image segmented by the proposed ANN model is much better than that obtained by the moment-preserving thresholding.<>