ISTA-Net Model-driven Deep Unfolding Network for Hyperspectral and Multispectral Image Fusion

Hyperspectral images (HSI) with high spatial and spectral resolutions have many applications in astronautics, re-mote sensing, and so on. However, it is challenging to obtain HSI with existing imaging techniques due to hardware limitations. In most cases, high-resolution multispectral (HrMS) images or low-resolution hyperspectral (LrHS) images are obtained. Therefore, the fusion of HrMS images and LrHS images for HSI super-resolution has attracted widespread attention. In this paper, we propose a network denoted as a model-based deep unfolding net-work(DuFNet) for hyperspectral image super-resolution (HSSR) task with clear interpretability. Specifically, we integrate the ISTA-Net into a well-established fusion network that is MHF-Net to fully take advantage of the generalization of the ISTA-Net. Experimental results demonstrate that the proposed NAM-DuFNet outperforms existing state-of-the-art methods in terms of subjective and objective results.