Fusion of Multi-band SAR Images Based on Nonsubsampled Contourlet and PCNN

Abstract

In this paper, a new fusion rule based on pulse coupled neural network (PCNN) and the clarity of images is proposed for multi-band synthetic aperture radar (SAR) images fusion. After the nonsubsampled Contourlet transform (NSCT) of the registered SAR images, we can get a flexible multiscale, multidirectional and shift-invariant representation of images. Then the linking strength of each neuron in PCNN is determined by the clarity of images, and new fire mapping images are obtained for each decomposed subband image taking part in the fusion. Experimental results show that the new fusion rule outperforms some classical rules, and after the inverse NSCT, the obtained fusion image can preserve much information of textures and edges of images than its counterparts.