|AI-based speckle suppression for SAR data|
|Organisatsiooni nimi||KappaZeta Ltd|
|Kokkuvõte||Synthetic aperture radar (SAR) is a powerful Earth Observation instrument providing continuous weather and sunlight independent measurements from the Earth. Due to the coherent nature of the instrument, SAR images are inherently speckled. There is multiplicative salt-and-pepper like noise, which is making the interpretation of SAR imagery more difficult. One cannot avoid speckle by cooling the instrument or improving the design, only possibility is to suppress speckle with post-processing the image data.|
Non-local means (Reference: Buades, A., Coll, B. and Morel, J.M., 2005, June. A non-local algorithm for image denoising. In 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05) (Vol. 2, pp. 60-65). IEEE.) is considered as one of the best conventional (non-AI) noise suppression methods for image data (including SAR data). The revolution of deep learning and AI is (besides other benefits) also providing powerful methods for noise suppression for image data.
The task is to test state of the art AI-based image denoising methods and compare them with non-local means to see if they are more effective in terms of noise suppression and preserving spatial resolution. It must be considered, how to design and implement AI-based speckle suppression methods so that the resulting data layers are physically correct, unbiased, and undistorted? This means preserving the phase information of the SAR data and the energy of the measurement (the sum of the pixels needs to be the same before and after denoising).
|Lõputöö kaitsmise aasta||2021-2022|
|Suhtlemiskeel(ed)||eesti keel, inglise keel|
|Nõuded kandideerijale||The candidates should have basic software development skills and know the fundamentals of machine learning and deep learning. Interest towards Earth Observation and satellite technology is required, but prior knowledge is not. KappaZeta can provide the basic training needed to get going.|
|Märksõnad||#machine_learning #deep_learning #earth_observation #satellite_data #noise_suppression #sar #synthetic_aperture_radar|