Several vision-inspired mathematical imaging problems such as image inpainting, segmentation and tracking have been modelled in the recent years by means of variational and geometrical tools weighting differently the orientations in the image at hand. Analytical approaches encode anisotropy in the choice of the functional spaces and nonlinear models, while geometrical techniques typically “lift” the ambient space in order to make the orientation one unknown of the problem. The aim of this workshop is to gather experts from both communities to promote discussions and collaborations. The validity of the models will be confirmed by their application to several imaging tasks.

Cortical-inspired functional lifting for image inpainting

Dario Prandi (CNRS - L2S, CentraleSupélec)

Computation of Curvature Penalized Shortest Paths via the Fast Marching Algorithm

Jean-Marie Mirebeau (Université Paris-Sud - CNRS - Université Paris-Saclay )

Anisotropic multiphase mean curvature flows with mobilities

Simon Masnou (Université Lyon 1)

A function space framework for structural total variation regularization with applications in inverse problems

Kostas Papafitsoros (Weierstrass Institute Berlin)

Organizers:

Luca Calatroni (CMAP, École Polytechnique CNRS)

Valentina Franceschi (INRIA Paris)

Dario Prandi (CNRS - L2S, CentraleSupélec)

Keywords:

anisotropic methods, computer vision, geometric control, image reconstruction, image segmentation, inverse problems, partial differential equation models