Pixel-level image fusion techniques in remote sensing: a review

Abstract

In the recent years remote sensing image fusion of satellite images has become a popular tool for analyzing different features presented on satellite images. We have analyzed only pixel-level image fusion techniques in this paper, which integrates a low-resolution multispectral (MS) image and high-resolution panchromatic (PAN) image to produce a more informative image than any of the single image. Generally, a PAN image is having a better spatial resolution, while the MS image is having a better spectral resolution than PAN image, due to this trade-off between MS and PAN image resolutions, it could be difficult to preserve spectral and spatial resolution in a single image without a fusion technique. In this paper, we have reviewed most popular and recent image fusion techniques and implemented them on Cartosat-1, RESOURCESAT-2, LANDSAT-8 data set. Results obtained from each method by visual analysis and quantitative measures indicated that UNB fusion algorithm outperforms other techniques used in this paper.