Generalized entropy mapping based neural network model and its application to image segmentation

A new neural network model for unsupervised pattern classification, which is known as generalized entropy mapping (GEM), is presented. The framework, characteristics and performance of generalized information entropic neural network are discussed. The GEM can be used for image segmentation in computer vision system. The global optimization net based on generalized entropy measure is given. The experimental results show that the performance of the GEM net is efficient in low-level visual information processing.