Jan Modersitzki: Research Interests

A generic task in modern image processing is image registration. The goal is to establish a correspondence between objects in two given images called reference and (moving) template, respectively. Particularly in medical imaging, registration is inevitable for example for motion correction and fusion of different image modalities. The following academical example illustrates the problem for two X-ray images of a human hand. For more information see my projects, my publications, and in particular my books Numerical Methods for Image Registration and FAIR: Flexible Algorithms for Image Registration.

Numerical analysis is the study of algorithms for the problems of continuous mathematics. My particular interest is on efficient and stable algorithms for large scale systems, including direct solvers and iterative schemes such as Krylov subspace or multigrid techniques.

My interest is on large scale optimization techniques for problems related to the discretization of partial differential equations. In particular I am interested in constrained problems as they arise, for example, in image processing.