Image reconstruction methods for high-resolution ground-based solar observations with deep learning
Solar observations from large-aperture ground-based telescopes face fundamental limitations that constrain our ability to resolve fine-scale structures in the solar atmosphere. Earths turbulent atmosphere significantly reduces the spatial resolution of those observations, requiring sophisticated post-facto image reconstruction methods.
This colloquium presents two novel deep learning techniques to address these challenges and to improve the image quality of these observations. We leverage architectures such as generative adversarial networks (GANs) and physics-informed neural networks (PINNs), which can enable robust image reconstruction and advance the study of the lower solar atmosphere with high spatial, spectral, and temporal resolution from the photosphere to the chromosphere.
Christoph Schirninger is a third year PhD student in astrophysics with a focus on solar physics and computer science at the University of Graz, Austria. His main research focuses on high-resolution solar image reconstruction methods for large-aperture ground-based telescopes using deep learning. Besides his main research, he has also been involved in various projects such as extending an AI tool to enhance observations and the intercalibration of instruments of space-based solar observations, 3D representations of the Sun from multi-viewpoint observations, solar irradiance estimations from extreme ultraviolet observations and solar flare detection and classification.