Liang et al. (2025) CSAN: A Channel–Spatial Attention-Based Network for Meteorological Satellite Image Super-Resolution

⚠️ Warning: This summary was generated from the abstract only, as the full text was not available.

Identification

• Journal: Remote Sensing
• Year: 2025
• Date: 2025-07-19
• Authors: Yuan Liu
• DOI: 10.3390/rs17142513

Research Groups

Not specified

Short Summary

The paper introduces CSAN, a channel-spatial attention-based network designed to super-resolve low-resolution spectral bands of meteorological satellite imagery to the highest available resolution.

Objective

• To alleviate the limitation of varying spatial resolutions across different radiative imaging bands in meteorological satellites by reconstructing missing spatial details in low-resolution (LR) bands.

Study Configuration

• Spatial Scale: Not specified (applicable to meteorological satellite imaging bands)
• Temporal Scale: Not specified

Methodology and Data

• Models used: Channel–Spatial Attention-based Network (CSAN), consisting of:
  • Information Fusion Unit: Adaptively fuses LR and HR images to capture inter-band spectral relationships.
  • Feature Extraction Module: Utilizes channel and spatial attention within a residual network.
  • Image Restoration Unit: Reconstructs high-resolution spatial details.
• Data sources: Meteorological satellite radiative imaging bands (visible to infrared).

Main Results

• The proposed CSAN network quantitatively and visually outperforms existing state-of-the-art super-resolution approaches.

Contributions

• Development of a novel deep learning architecture (CSAN) that integrates channel and spatial attention to effectively leverage the relationship between different spectral bands for image super-resolution.

Funding

Not specified

Citation

@article{Liang2025CSAN,
  author = {Liang, Weiliang and Liu, Yuan},
  title = {CSAN: A Channel–Spatial Attention-Based Network for Meteorological Satellite Image Super-Resolution},
  journal = {Remote Sensing},
  year = {2025},
  doi = {10.3390/rs17142513},
  url = {https://doi.org/10.3390/rs17142513}
}