Multispectral and hyperspectral image fusion with spatial-spectral sparse representation

Fusing a high spatial resolution multispectral image (HR-MSI) with a low spatial resolution hyperspectral image (LR-HSI) of the same scenario to acquire a high spatial resolution hyperspectral image (HR-HSI) has recently attracted more and more attention. We propose a novel spatial-spectral sparse representation (SSSR) based approach for the fusion of an HR-MSI and an LR-HSI of the same scenario in this paper. In the proposed SSSR method, we formulate the fusion problem as the estimation of spectral basis and coefficients from the LR-HSI and HR-MSI. To better model the spatial and spectral characteristics of the HR-HSI, we incorporate the non-local spatial similarities, priors of the spectral unmixing, and a sparse prior to the fusion problem. Meanwhile, instead of keeping the spectral basis fixed, we design the alternative optimization algorithm for the estimation of spectral basis and coefficients, which can achieve the accurate reconstruction. Experimental results on both non-blind fusion and blind fusion cases demonstrate the effectiveness of the SSSR approach.