The medial feature detector: Stable regions from image boundaries

Avrithis, Y; Rapantzikos, K

dc.contributor.author	Avrithis, Y	en
dc.contributor.author	Rapantzikos, K	en
dc.date.accessioned	2014-03-01T02:53:30Z
dc.date.available	2014-03-01T02:53:30Z
dc.date.issued	2011	en
dc.identifier.uri	https://dspace.lib.ntua.gr/xmlui/handle/123456789/36362
dc.subject.other	Computational requirements	en
dc.subject.other	Duality properties	en
dc.subject.other	Euclidean spaces	en
dc.subject.other	Feature detector	en
dc.subject.other	Group marching	en
dc.subject.other	Image boundaries	en
dc.subject.other	Image gradients	en
dc.subject.other	Image Structures	en
dc.subject.other	Linear-time algorithms	en
dc.subject.other	Local feature detectors	en
dc.subject.other	Marching method	en
dc.subject.other	Medial axis	en
dc.subject.other	Reduced memory	en
dc.subject.other	Saddle point	en
dc.subject.other	Scale spaces	en
dc.subject.other	Stable region	en
dc.subject.other	State-of-the-art performance	en
dc.subject.other	Voronoi	en
dc.subject.other	Weighted distance	en
dc.subject.other	Clustering algorithms	en
dc.subject.other	Image processing	en
dc.subject.other	Feature extraction	en
dc.title	The medial feature detector: Stable regions from image boundaries	en
heal.type	conferenceItem	en
heal.identifier.primary	10.1109/ICCV.2011.6126436	en
heal.identifier.secondary	http://dx.doi.org/10.1109/ICCV.2011.6126436	en
heal.identifier.secondary	6126436	en
heal.publicationDate	2011	en
heal.abstract	We present a local feature detector that is able to detect regions of arbitrary scale and shape, without scale space construction. We compute a weighted distance map on image gradient, using our exact linear-time algorithm, a variant of group marching for Euclidean space. We find the weighted medial axis by extending residues, typically used in Voronoi skeletons. We decompose the medial axis into a graph representing image structure in terms of peaks and saddle points. A duality property enables reconstruction of regions using the same marching method. We greedily group regions taking both contrast and shape into account. On the way, we select regions according to our shape fragmentation factor, favoring those well enclosed by boundaries-even incomplete. We achieve state of the art performance in matching and retrieval experiments with reduced memory and computational requirements. © 2011 IEEE.	en
heal.journalName	Proceedings of the IEEE International Conference on Computer Vision	en
dc.identifier.doi	10.1109/ICCV.2011.6126436	en
dc.identifier.spage	1724	en
dc.identifier.epage	1731	en