Tuesday, September 26
Minisymposium 12: Optimization methods for inverse problems in imaging
Time: 11:00 - 13:00
Room: L1.202, Building L
Organiser: Elena Loli Piccolomini, University of Bologna, and Luca Zanni, University of Modena and Reggio Emilia
Many imaging applications, such as denoising, deblurring, X-ray CT Tomography or Magnetic Resonance Imaging, segmentation and super-resolution are described by continuous ill-posed inverse models, leading to ill-conditioned discrete problems. Their solution is computed by solving large-scale constrained or unconstrained optimization problems involving a data fidelity function and one or more regularization terms. For handling these regularized problems, many interesting optimization strategies have been developed in the last years, which have given rise to effective image reconstruction tools.
This minisymposium aims to strengthen the interaction between the imaging and the optimization communities, highlighting the successful results obtained so far and indicating topics on in which more work is needed.
|11:00 - 11:30||Jack Spencer (University of Liverpool)|
On optimization of a network of minimal paths for 3D image segmentation
|11:30 - 12:00||Germana Landi (University of Bologna)|
Automatic adaptive multi-penalty regularization for linear inverse problems
|12:00 - 12:30||Jean-Christophe Pesquet (University Paris-Saclay)|
A variational Bayesian approach for image restoration with Poisson-Gaussian noise
|12:30 - 13:00||Simone Rebegoldi (University of Modena and Reggio Emilia)|
A block-coordinate variable metric line-search based proximal-gradient method for nonconvex optimization