RoIDE article is publised in IEEE Transactions on Geoscience and Remote Sensing
๐ Happy Lunar New Year to My Dear Colleagues! ๐
Wishing everyone happiness, prosperity, and success in the year ahead! ๐ฎ๐โจ
Excited to start this year with new publication and an open-source contribution! ๐ป
๐ Solving a Major Challenge in Satellite Imaging! ๐ฐ๏ธ
One of the biggest challenges in satellite imaging is handling extreme variations in brightness within a single frame. ๐
๐ The Problem:
Satellite images often contain regions that are too dark (shaded areas, forests, water bodies) or too bright (snow, clouds, deserts) due to environmental conditions. Traditional enhancement methods often overexpose bright areas while trying to enhance darker regions, leading to loss of crucial details.
๐ก Our Solution: RoIDE (Region of Interest-Focused Dynamic Enhancement)
Our latest research, published in IEEE Transactions on Geoscience and Remote Sensing, sponsored by IEEE Geoscience and Remote Sensing Society (GRSS), introduces RoIDE, a deep learning-based enhancement framework that:
- โ Generates multi-exposure images from a single satellite image
- โ Dynamically fuses exposures to enhance regions of interest (RoI)
- โ Preserves details in both dark and bright areas without overexposure
๐ Read the full paper:
๐ DOI: 10.5281/zenodo.14792938
๐ป Open-Source Code Now Available!
To support the research community, Iโve made our RoIDE implementation open-source:
๐ My GitHub Repository: https://lnkd.in/gsEAY3Ap
๐ฅ Dataset: Available on Zenodo with DOI: 10.5281/zenodo.14792938 https://lnkd.in/gn-z62mg
Why This Matters
From disaster response to urban planning, having clearer and more balanced satellite images can drastically improve decision-making. Our approach allows for better object detection, climate analysis, and environmental monitoring without introducing artifacts or distortions.
Special Thanks ๐
This research wouldnโt have been possible without the support and collaboration of my amazing colleagues from:
๐ National Taipei University of Technology, Taiwan
๐ National Taiwan University of Science and Technology, Taiwan
๐ VILNIUS TECH - Vilnius Gediminas Technical University, Lithuania
Your contributions and partnership have been invaluable! Looking forward to more exciting collaborations ahead. ๐
Join the Discussion!
Iโd love to hear your thoughts! Whether you work in remote sensing, AI, or image processing, letโs collaborate on future advancements. ๐
๐ง Contact me:
Andrew (Trong-An) Bui
๐ Institute of Aerospace and System Engineering
๐ National Taipei University of Technology, Taiwan
If you find this work useful, please star โญ the repository, try it out, and share your thoughts!
#DeepLearning #SatelliteImaging #RemoteSensing #AI #GeospatialAnalysis #OpenSource
