A Parallel Fusion Algorithm of Remote Sensing Images

The fusion of remote sensing images can effectively combine two or more source images into a single image with more information of the spatial resolution and spectral information. In this paper we present a novel parallel fusion algorithm based on intensity-hue-saturation (IHS) transform and nonsubsampled contourlet transform (NSCT) to deal with the panchromatic (PAN) image and multispectral (MS) image. The rule based on regional gradient is applied to the combination of low-frequency coefficients of NSCT, and the rule based on weight is used for the combination of high-frequency coefficients of NSCT. In addition, we investigate some factors that affect the performance of our approach. Experimental results show that our algorithm outperforms principal component analysis (PCA), IHS, wavelet transform and IHS transform-based (WT-IHS), and contourlet transform and IHS transform-based (CT- IHS). Our algorithm has good scalability simultaneously.