Images are most often represented by pixel data on Cartesian
grids. However, finite element models may be preferred in
certain situations. Examples include images on triangulated
surfaces, or when higher-order representations and adaptivity
play a role. Choosing a finite element discretization has
interesting implications on the algorithmic solution of
image restoration and related problems as well as their
duals, which will be highlighted by the speakers in this
minisymposium.
- Discrete total variation with finite elements
- Roland Herzog (Technische Universität Chemnitz)
- Adaptive finite element approximation of the ROF model
- Marijo Milicevic (University of Freiburg)
- Fast and robust boundary segmentation using 2nd order shape sensitivity of variational models
- Gunay Dogan (Theiss Research, NIST)
- Adaptive finite elements for Mumford-Shah-type functionals in transport network modelling
- Rossmanith Carolin (Westfälische Wilhelms-Universität Münster)
- Organizers:
-
Roland Herzog (Technische Universität Chemnitz)
-
Stephan Schmidt (University of Würzburg)
- Keywords:
- finite elements, image reconstruction, nonlinear optimization, numerical linear algebra