SSMEUN: Spatial-Spectral Mamba Enhanced Unfolding Network for Pan-Sharpening

Pan-sharpening aims to fuse panchromatic (PAN) images and low-resolution multispectral (LR-MS) images. Although recent pan-sharpening methods based on deep unfolding networks (DUNs) improve the model’s interpretability, they do not fully explore the relationship between LR-MS and PAN images. To tackle the issue, we propose a spatial-spectral Mamba-enhanced unfolding network (SSMEUN), which not only improves the model’s interpretability but also enhances the exploration of relationships. Specifically, SSMEUN uses a data module to encode the degradation information and a prior module to model relationships. Additionally, it introduces a Spectral-Redistribution Mechanism to address uneven band-wise information distribution in LR-MS images, and a Multi-directional Flattening Mechanism to alleviate contextual loss from unidirectional modeling in the data module. Experiments validate the efficiency and accuracy of our method.