Multiclass segmentation of SAR image using modified unit-linking pulse coupled neural network

A method for segmentation of SAR images based on modified unit-linking pulse coupled neural networks (unit-linking PCNN) is presented. The segmentation images using traditional unit-linking PCNN are binary, and we modify unit-linking PCNN to be two levels in order to make it segment images for more classes. The primary level corresponds to finding the clustering centers, and the similar neurons are captured using unit-linking PCNN in the secondary level. Because the grey distribution of SAR image is uneven, the gray mean of the neuron's n times n window image is used as the input pulse signal. Experimental results show that the proposed method is effective.