Faculty of Informatics Vienna University of Technology Institute of Computer Aided Automation PRIP Home PRIP Home
Personal tools
You are here: Home Teaching PRIP-Talks Fabian Hollaus: Automated inpainting of unknown palimpsest regions

Fabian Hollaus: Automated inpainting of unknown palimpsest regions

Spezifikationsvortrag DA

  • Presentation
When Mar 04, 2010
from 02:15 pm to 02:15 pm
Where Sem 183/2
Add event to calendar vCal
The goal of image inpainting is to restore certain parts of an image, in such a manner, that a viewer can not detect the restored parts. The region that should be inpainted is usually defined by a binary mask and only this region is altered during the inpainting process. In this thesis the main focus lies on the inpainting of palimpsests. The goal is to restore the underlying, older text and therefore the binary mask - that marks the inpainting region - is the overlying, younger text.
In this diploma thesis several inpainting techniques will be compared. The inpainting methods will be evaluated on palimpsests and natural images (which can be encountered in real life). Most of the effort will be spent on an algorithm that is called the Fields of Experts (FoE) - framework. This framework is based on Bayesian inference and allows the learning of high order Markov Random Fields priors. The inpainting task is performed with the help of such a learned prior. The FoE inpainting technique is based on probability theory and will be compared to other state of the art algorithms which typically rely on image gradients.
Document Actions