Multi-Scale and Multi-Stream Fusion Network for Pansharpening

Pansharpening refers to the use of a panchromatic image improve spatial resolution multi-spectral while preserving spectral signatures. However, existing pansharpening methods are still unsatisfactory at balancing trade-off between enhancement and fidelity. In this paper, multi-scale multi-stream fusion network (named MMFN) that leverages information source images is proposed. The proposed architecture simple, yet effective, can fully extract various spatial/spectral features different levels. A multi-stage reconstruction loss was adopted recover pansharpened in each block, which facilitates stabilizes training process. qualitative quantitative assessment on three real remote sensing datasets (i.e., QuickBird, Pléiades, WorldView-2) demonstrates approach outperforms state-of-the-art methods.