Aarhus University Seal / Aarhus Universitets segl

Lukas Esterle

Indexing digital images based on overlap

Publikation: Bidrag til tidsskrift/Konferencebidrag i tidsskrift /Bidrag til avisKonferenceartikelForskningpeer review

DOI

Image retrieval currently relies heavily on metadata and present search engines do not exploit spatial relations between images. Therefore searching for images of a certain location manually is a time consuming task. Typically there are pair wise spatial relations in an unsorted set of photos taken at the same location from different angles and different viewpoints. By identifying and indexing spatial relations of different photos an image search engine can provide additional results based on these relationships. This paper presents a novel approach to identify pair wise spatial relations between images and a method to index and search through the identified spatial relations. We further describe an application, which implements this robust and surprisingly simple technique.

OriginalsprogEngelsk
TidsskriftProceedings of the 2011 5th FTRA International Conference on Multimedia and Ubiquitous Engineering, MUE 2011
Sider (fra-til)134-137
Antal sider4
DOI
StatusUdgivet - 14 sep. 2011
Eksternt udgivetJa
Begivenhed2011 5th FTRA International Conference on Multimedia and Ubiquitous Engineering, MUE 2011 - Loutraki, Grækenland
Varighed: 28 jun. 201130 jun. 2011

Konference

Konference2011 5th FTRA International Conference on Multimedia and Ubiquitous Engineering, MUE 2011
LandGrækenland
ByLoutraki
Periode28/06/201130/06/2011

Se relationer på Aarhus Universitet Citationsformater

ID: 170586014