A pan-sharpening model using dual-branch attention-guided diffusion networks

Huangqimei Zheng, Chengyi Pan, Xin Jin, Michal Wozniak, Puming Wang, Shin-Jye Lee, Qian Jiang

Abstract: Highlights•The novel dual-branch pan-sharpening model DADiff fuses PAN and MS images.•Diffusion module captures global joint distributions through progressive noise removal.•Attention-guided branch enhances spatial and spectral features via multi-scale attention.•Experiments demonstrate DADiff’s superior spatial and spectral fidelity across multiple datasets.