IACM Colloquium
Speaker:
Kostas Papafitsoros
Title:
Total variation-based image reconstruction methods
Abstract:
In this talk, we will initially give an introduction to image reconstruction methods with variational techniques using total variation-based regularisation functionals of first and second order. We will examine the structure of solutions of the corresponding minimisation problems from a theoretical point of view and we will discuss the advantages but also the limitations on the quality of the reconstructed images. We will analyse the roles of the regularisation parameters and we will introduce a bilevel minimisation scheme that is able to automatically compute these parameters. The computed parameters are also spatially distributed resulting in better reconstruction of the fine scale details of the image. We will present numerical results in image denoising and magnetic resonance imaging (MRI).
Short Bio:
Kostas Papafitsoros received a 5-year diploma degree from the School of Applied Mathematics and Physical Sciences, National Technical University of Athens, in 2007. He holds two M.Sc. degrees, one in Pure Mathematics from the University of Athens in 2009 and one in Mathematical Modelling and Scientific Computing from the University of Oxford in 2010. He received his Ph.D. degree in 2014 from the University of Cambridge where he also stayed as an EPSRC Doctoral Prize fellow at the Department of Applied Mathematics and Theoretical Physics until April 2015. He was an Alexander von Humboldt Foundation postdoctoral fellow at the Institute for Mathematics, Humboldt University of Berlin and at the Weierstrass Institute for Applied Analysis and Stochastics (WIAS) from 2015 until 2017. He is currently a research scientist at WIAS at the group ''Nonsmooth Variational Problems and Operator equations''. His research interests lie in the area of mathematical imaging, variational methods, optimisation, machine learning as well as application of mathematics to environmental conservation projects.
Time, Date & Location:
15:00, Tuesday 3rd May 2022 @ Orfanoudakis Room (main building)
Zoom Info:
https://us06web.zoom.us/j/7761040361?pwd=dEs4TFZxck01Vk5PQWR1TUxKdndwUT09
Meeting ID: 776 104 0361
Passcode: 787479