Transport metrics in image and shape processing
Organizers:
Dirk Lorenz (Technical University Braunschweig, Germany) and Benedikt
Wirth (University of Muenster, Germany)
Abstract:
During recent years the use of the Wasserstein distance and its variants have become
popular in variational image processing and shape analysis. They are employed, for
example, as data or delity term, as regularizer or to introduce a geometry on the space
of shapes. Among their advantages, they provide a natural description of transport
and material
ux and are applicable to input data of low regularity such as Radon
measures. Furthermore, they can be formulated in diverse ways, for instance based on
ow elds or on transport plans in product spaces, and the dierent formulations have
their own merit in applications. This minisymposium is aimed at bringing together
experts on different versions and applications of transport-based metrics in image and
shape processing in order to exchange recent successful modeling and computational
approaches between the different areas.
|