Satellite Image Fusion Using Maximization of Non-Gaussianity?

Image fusion is a technique for combining images from different sources to obtain a single image with enhanced information content. This paper proposes an image fusion method to merge panchromatic (PAN) and multispectral (MS) remote sensing satellite images using genetic algorithm to maximize the nongaussianity of the independent components of ICA. The genetic algorithm evolves the mixing matrix of the independent components of the MS image by maximizing the kurtosis. The proposed method is applied to Quickbird, Ikonos, and Worldview satellite image data. Performance evaluation of the proposed method is compared with that of IHS, PCA, and ICA based image fusion methods. Experimental results show optimum performance of the proposed method in terms of spatial resolution and color preservation of the fused images with the three different types of satellite image data.