Hyperspectral image fusion by hybrid regularizations with local low-rank

Hyperspectral images usually have higher spectral resolution but lower spatial resolution, compared with the multispectral images. Low spatial resolution brings difficulties to the practical applications of hyperspectral images. Therefore, to get high spatial resolution hyperspectral image, it is very important to fuse low spatial resolution hyperspectral image with the high spatial resolution multispectral image in the same scene. In this paper, we propose a hybrid regularization model by integrating sparse prior, local low-rank regularization and total variation based on l 2 norm to reconstruct high spatial resolution hyperspectral images. In addition, we design an alternating direction method of multipliers (ADMM) to solve it. The experimental results show the superiority and competitiveness of our method over the state-of-the-art methods.