Imaging Seminar - Seeing the unseen: computational imaging across applications

Abstract: Computational imaging combines mathematical modelling, inverse problems, optimisation, and machine learning to recover information from indirect and noisy measurements. In this seminar, I will present recent research on physics-informed reconstruction methods and learning-based approaches, illustrating how a common mathematical framework can address diverse imaging challenges. Applications will include reconstruction of ancient frescoes in cultural heritage, computational models of vision for understanding geometric optical illusions, and advanced reconstruction procedures for image deconvolution/super-resolution in advanced fluorescence microscopy

Bio: Prof. Luca Calatroni is an Associate Professor at the University of Genoa (DIBRIS), where he leads the Computational Imaging & Learning unit at the Machine Learning Genoa Center (MaLGa). He received his PhD in applied mathematics from the University of Cambridge in 2015. He then held postdoctoral positions in Genoa and at the École Polytechnique, and was a CNRS researcher in Sophia-Antipolis from 2019 to 2024. He is the PI of the ERC Starting Grant MALIN. His research is in mathematical imaging and inverse problems, at the meeting point of variational methods, optimisation and machine learning. He is especially interested in learning-based reconstruction methods that stay interpretable and reliable because they keep the physics of the acquisition process in the model. His work ranges from biomedical imaging, where he pushes the resolution limits of fluorescence microscopy and medical scans, to mathematical models of human vision and visual illusions, and to the digital restoration of damaged artworks and ancient manuscripts.