Pansharpening Multispectral Images Based on Unconstrained Least Square Spectral Unmixing

A new fusion framework for pansharpening multi spectral (MS) image is suggested in this paper. The introduced method relies on linear spectral unmixing, and employs unconstrained least square (ULS) estimation for combining the high spatial features of the panchromatic (PAN) image and the rich spectral features of the MS image (MSI), in one single high spatial-spectral image. Applied to real coincident MS and PAN data sets with different texture complexity, the proposed approach can fuse the spatial and the spectral characteristics in a single image with the minimum of spectral distortion in comparison to some famous methods. Moreover, incorporating the ULS in the unmixing process is quite simple to formulate and straightforward to implement.