A Study on Transformer-Based Super-Resolution Network Design Techniques Utilizing Low Earth Orbit Multispectral Satellite Images
Abstract
Low Earth orbit satellites play a vital role in Earth environment monitoring and industrial applications through multispectral data. However, the limited resolution poses challenges for applications that require fine details and for real-time processing. To address these issues, this study proposes a transformer-based super-resolution deep learning network that leverages the multi-channel characteristics of multispectral imagery and captures inter-channel interactions. The proposed network and procedure were validated in terms of data processing efficiency, reconstruction performance, and processing speed, demonstrating their effectiveness.
Fields of Study & Subjects
Advanced Image Processing Techniques
Satellite Image Processing and Photogrammetry
Remote-Sensing Image Classification
Publication details
- Journal / source
- Journal of Space Technology and Applications
- Volume
- 6
- Issue
- 1
- Pages
- 24–37
- Publication date
- 1 February 2026
- Language
- EN