Advanced Search

Browse by Discipline

Scientific Societies

E-print Alerts

Add E-prints

E-print Network

  Advanced Search  

Lehrstuhl fr Informatik 8 Zeit: Donnerstag, 30.04., 16.30 Uhr

Summary: Lehrstuhl für Informatik 8
Zeit: Donnerstag, 30.04., 16.30 Uhr
Ort: AH I, Ahornstr. 55
Referent: Herr Prof. Dr. Daniel Cremers
Universität Bonn
Titel: Optimal Solutions for Spatially Continuous
Labelling Problems
Numerous computer vision problems can be cast as labelling problems where each
point is assigned one of several labels. The case of two labels includes problems like
binary segmentation and multi-view reconstruction. The case of multiple labels in-
cludes problems such as stereo depth reconstruction and image denoising. In my
presentation, I will introduce methods of convex relaxation and functional lifting which
allow to optimally solve such labelling problems in a spatially continuous setting. Ex-
perimental results demonstrate that these spatially continuous approaches provide
numerous advantages over spatially discrete (graph cut) formulations, in particular
they are easily parallelized (lower runtime), they require less memory (higher resolu-
tion) and they do not suffer from metrication errors (better accuracy).


Source: Ábrahám, Erika - Fachgruppe Informatik, Rheinisch Westfälische Technische Hochschule Aachen (RWTH)


Collections: Computer Technologies and Information Sciences