Remote sensing image fusion using best bases sparse representation

A new technique based on best bases sparse representation is proposed for fusion of remote sensing images. In order to carry out multi-resolution image fusion, low-resolution image is upscaled to match resolution of high-resolution images. Corresponding patches from remote sensing images are represented by finding out best bases from over-complete dictionaries comprising of elements derived from basis function of DCT, Wavelets, ridgelets, and curvelets. The corresponding bases of image patches are combined based on local information parameter (LIP) derived from respective patches. The use of LIP helps ensure transfer of details in high-resolution image into fused image.