Howarth, Peter and Rüger, Stefan
(2005).
|
Due to copyright restrictions, this file is not available for public download |
| URL: | http://www-gsi.dec.usc.es/ecir05/ |
|---|---|
| DOI (Digital Object Identifier) Link: | https://doi.org/10.1007/978-3-540-31865-1_32 |
| Google Scholar: | Look up in Google Scholar |
Abstract
We have applied the concept of fractional distance measures, proposed by Aggarwal et al. [1], to content-based image retrieval. Our experiments show that retrieval performances of these measures consistently outperform the more usual Manhattan and Euclidean distance metrics when used with a wide range of high-dimensional visual features. We used the parameters learnt from a Corel dataset on a variety of different collections, including the TRECVID 2003 and ImageCLEF 2004 datasets. We found that the specific optimum parameters varied but the general performance increase was consistent across all 3 collections. To squeeze the last bit of performance out of a system it would be necessary to train a distance measure for a specific collection. However, a fractional distance measure with parameter p = 0:5 will consistently outperform both L1 and L2 norms.
| Item Type: | Conference or Workshop Item |
|---|---|
| Copyright Holders: | 2005 Springer-Verlag |
| ISSN: | 0302-9743 |
| Extra Information: | Published in: D.E. Losada and J.M. Fernandez-Luna (Eds.): ECIR 2005, LNCS 3408, pp. 447–456, 2005 |
| Academic Unit/School: | Faculty of Science, Technology, Engineering and Mathematics (STEM) > Knowledge Media Institute (KMi) Faculty of Science, Technology, Engineering and Mathematics (STEM) |
| Item ID: | 29878 |
| Depositing User: | Stefan Rüger |
| Date Deposited: | 27 Oct 2011 14:28 |
| Last Modified: | 04 Oct 2016 19:46 |
| URI: | http://oro.open.ac.uk/id/eprint/29878 |
| Share this page: |     |
Altmetrics | Scopus Citations |