Thursday, September 28

Minisymposium 14: Non-smooth optimization and computational 3D image processing

Time: 8:00 - 10:00
Room: L1.201, Building L
Organiser: Stephan Schmidt, Universität Würzburg, and Jose Vidal Nuñez, Technische Universität Chemnitz

Chair: TBA

With the rapid growth of 3D scanning and sophisticated medical imaging techniques, there is now a growing necessity to process textured 3D objects. In addition to sampling errors of the imaging equipment, lossy compression in streaming geometry and textures for internet applications and eCommerce are also a potential source of noise necessitating post production. Another field where the texture is of great importance is neuroimaging, where random noise enters due to eddy-current distortions or physical motion during the magnetic resonance imaging (MRI) process.

The aim of this mini-symposium is to bridge the gap between processing, e.g. denoising, flat images and surfaces fairing, allowing to treat real world objects holistically by including both geometric and texture information. Feature preservation, for example via norms based on total variation, links to non-smooth optimization and computational geometry.

Speakers:

8:00 - 8:30Tuomo Valkonen (University of Liverpool)
Block-proximal methods with spatially adapted acceleration
8:30 - 9:00Kristian Bredies (Universität Graz)
Accelerated and preconditioned Douglas-Rachford algorithms for the solution of variational imaging problems
9:00 - 9:30Marc Herrmann (Universität Würzburg)
Total variation image reconstruction on smooth surfaces
9:30 - 10:00Ferdia Sherry (University of Cambridge)
Learning the sampling pattern for MRI