The source code of paper "Rethinking Pan-sharpening via Spectral-band Modulation".
The core innovation of our paper is the Spatially-Adaptive Spectral Modulation Network (SSMNet) that pays special attention to the individual modulation of every spectral band conditioned on the PAN image to achieve superior pan-sharpening results. SSMNet comprises three key components: the Source-Aware Spectral Modulator for local band-specific information, Cross-Band Information Aggregation for cohesive spectral representation, and Cross-Stage Feature Integration for the final image output. The paper's results demonstrate SSMNet's effectiveness, outperforming state-of-the-art methods across various datasets. For more detailed insights, refer to the paper linked above.
Model Architecture:
If you use any ideas from the paper or code from this repo, please consider citing:
@article{liu2023rethinking,
title={Rethinking Pan-Sharpening via Spectral-Band Modulation},
author={Liu, Xinyang and Hou, Junming and Cong, XiaoFeng and Shen, Hao and Lou, Zhuochen and Deng, Liang-Jian and You, Jian Wei},
journal={IEEE Transactions on Geoscience and Remote Sensing},
year={2023},
publisher={IEEE}
}