More Effective Earth Observation for Climate Action Through Learned Data Compression in CubeSats
- Project and Master Subjects 2025-2026
-
Past Projects
-
Project and Master Subjects 2024-2025
- Improving Images for Climate Action
- Every Variable Everywhere All at Once
- Protecting Water Resources through Machine Learning and Hyperspectral Imaging in Remote Sensing CubeSats
- More Effective Earth Observation for Climate Action Through Learned Data Compression in CubeSats
- Mitigating Camera Artifacts in HYPSO Data for Improved Climate Monitoring
- Characterization of High-resolution Spectral Imager
- A Foundational Unmixing Model for the HYPSO satellites
- Hyper/Multispectral image fusion with HYPSO-2
- Game theory applied to energy optimal satellite attitude control
- Mu-analysis for agile satellite attitude control maneuvers
- Randomized optimization applied to super-agile satellite operations
- Modelling super-agile satellite operations for optimization
- Enabling high-accuracy HYPSO image georeferencing by high-accuracy satellite pose estimation through postprocessing of satelitte sensor data
- High-accuracy attitude determination of Earth observation satellites
- Agile Earth Observation Satellite simulation studies
- Multi-angle image analysis and what we can learn about the atmosphere
- GNSS-R: Simulator design of a GNSS-Reflectometry small satellite
- GNSS-R: GNSS jamming and spoofing source localization from a small satellite
- GNSS-R: Maritime Surveillance using GNSS-Reflectometry
- Project and Master Subjects 2023-2024
- Project and master assignments 2022
-
Project and Master Subjects 2024-2025
More Effective Earth Observation for Climate Action Through Learned Data Compression in CubeSats
Project Focus
The aim of this project is to develop models that compresses HYPSO data using modern machine learning principles. Part of the project will attempt to compare the different types of models with existing standards for image compression in space. These standards are known as CCSD121 and CCSDS123.
Impact
Space technology plays a crucial role in achieving various Sustainable Development Goals (SDGs) set by the UN. The NTNU SmallSat Lab's HYPSO satellites, launched in 2022 and with a successor planned, utilize hyperspectral imagers to capture detailed information beyond the visible. This data allows us to detect and monitor water bodies like oceans, fjords, and lakes, including vital yet potentially harmful algae. The HYPSO satellites also contribute to climate change studies by imaging the Arctic region. Ultimately, HYPSO data aims play a role in achieving Climate Action, Preserving Life Below Water, and ensure access to Clean Water and Sanitation.
Tasks and Expected Outcomes
Collect a robust dataset of hyperspectral imagery from the HYPSO CubeSats. Review the current literature on end-to-end compression. Select apt models for testing and evaluation for desktop computers. Write a report of your findings.
Who We Are Looking For
We are seeking a highly motivated final year student in Cybernetics, Electronics, or a related field with an interest in image processing and remote sensing applications. Experience with signal processing techniques is not mandatory. The project will be adapted to the student's background and goals.
How we work
At the NTNU SmallSat Lab we encourage collaboration and try to get our group to help each other. To facilitate this, we encourage the students to meet us in the lab as well as arrange common lunches and workshops where the students and supervisors can learn from each other.
For further questions please contact Sivert Bakken
https://www.ntnu.no/ansatte/sivert.bakken
Relevant References
https://www.tensorflow.org/tutorials/generative/data_compression
https://doi.org/10.48550/arXiv.1611.01704
https://doi.org/10.1016/j.sigpro.2021.108346
https://arxiv.org/abs/2011.03029