Hyperspectral image fusion with a new hybrid regularization

Hyperspectral images usually have higher spectral resolution than the multispectral images. However, the spatial resolution of hyperspectral images is lower, which limits their practical applications. Therefore, to obtain high spatial resolution hyperspectral image (HR-HSI), it is very important to fuse a low spatial resolution hyperspectral image with a high spatial resolution multispectral image in the same scene. In this paper, we propose a new sparse hyperspectral image fusion model. To better model the spatial and spectral characteristics of the HR-HSI, we incorporate a sparse prior, the local low-rank regularization and the total variation based on ℓ _1 norm. To solve the problem efficiently, we design an alternating direction method of multipliers (ADMM). The experimental results show the effectiveness and competitiveness of our method over the state-of-the-art methods.