Multiscale image fusion for pansharpening of multispectral images using saliency detection

Abstract

Pansharpening is a technique used for fusion of low spatial resolution multispectral images with high spatial resolution panchromatic image. The objective of pansharpening is to enhance the spatial resolution of multispectral image while preserving its spectral information. The commonly used pansharpening methods like IHS, PCA, Gram Schmidt and Wavelet based method compromise either on spatial or spectral resolution. This paper presents a modified multiresolution analysis based pansharpening technique that is based on saliency detection of an image. This method fuses the spatial and spectral information of the images independently; therefore the fused image is obtained with high spatial as well as spectral information. The technique has been implemented on different datasets of multispectral and panchromatic images using different sets of fusion rules. The results reveal that the modified saliency detection based pansharpening technique performs better than the existing pansharpening technique and other commonly used pansharpening techniques.