Faculty of Informatics Vienna University of Technology Institute of Computer Aided Automation PRIP Home PRIP Home
Personal tools
You are here: Home People Julian Stöttinger Publications Scale Invariant Dissociated Dipoles

Scale Invariant Dissociated Dipoles

In this paper we investigate the use of dissociated information to represent image structures in a scale invariant way. Our contribution consists in an extension to the dissociated dipoles based descriptor that has proven to be as robust to image transformations as SIFT, while containing 6 times less data. We demonstrate that our extension has better performance when scaling of images takes place. Further, we show the benefits of the more stable color interest points for both feature localization and scale selection. We are able to demonstrate that colored scale invariant dipoles have comparable or better matching scores than SIFT in an exact nearest neighbor matching scheme.

2010oeagm_marco.pdf — PDF document, 635Kb