Bayesian fusion of multispectral and panchromatic images

In this paper, a model-based pansharpening method based on the super-resolution (SR) paradigm is developed. The spatial and spectral relationships between the desired high resolution multispectral (MS) image and the observed low resolution MS and panchromatic (Pan) images are combined with an appropriate image prior model via Bayesian theory. Maximum a posteriori (MAP) estimation is employed to convert the inverse problem of restoring the desired MS image into a constrained optimization problem. The final desired MS image is obtained by the conjugate gradient (CG) algorithm. Experimental results on two datasets show the effectiveness of the proposed method compared to the well-known pansharpening methods according to the quantitative evaluation metrics and visual inspection.