A new neural network model based approach to unsupervised image segmentation

This paper proposes a new neural network model UMAN to perform unsupervised image segmentation. In the neural net,the generalized information entropy is used as the quantitative description and measurement of the system stability and asymptotication, and the disadvantage of generalized energy function is avoided also. The improved Kohonen non-linear mapping structure not only enhances the clustering features,but also reduces the redundant information. In the network,the internal layer and node number are determined dynamically by the system. The interaction and a prior knowledge are not required. The unsupervised self - learning function expresses the characteristics of the low level visual information processing. The UMAN model could process various types of images and is with strong adaptability. Experimental result shows that the model and its algorithm is efficient,practical and robust.