Skip to Main content Skip to Navigation
Reports

About optimal use of color points of interest for content-based image retrieval

Abstract : In content-based image retrieval systems, the main approaches based on the query-by-example paradigm involve an approximate search on the whole image, which requires a global description of it. When considering tasks like object recognition or partial queries on particular area, these methods become inadequate and more local characterizations must be employed. In this context, image description based on points of interest appear best adapted. The point characterization which proved reliable is based on invariants to rotation and in particular on combinations of the Hilbert's differential invariants. For gray value images, such a description used to be considered up to third order at least. More recently, generalizations to color images were proposed for stereovision and image retrieval. Some of them propose to consider the invariants only at first order and to enrich the characterization with geometrical constraints for describing spatial relations between points, while others consider higher order invariants and compute some combinations of them to achieve illumination changes invariance. In this report, we discuss the advantages and drawbacks of these different choices, with the aim of proposing an optimal use of color points of interest for content-based image indexing and retrieval.
Document type :
Reports
Complete list of metadata

https://hal.inria.fr/inria-00072149
Contributor : Rapport de Recherche Inria <>
Submitted on : Tuesday, May 23, 2006 - 7:54:50 PM
Last modification on : Friday, April 24, 2020 - 8:48:07 AM
Long-term archiving on: : Sunday, April 4, 2010 - 10:55:38 PM

Identifiers

  • HAL Id : inria-00072149, version 1

Collections

Citation

Valérie Gouet, Nozha Boujemaa. About optimal use of color points of interest for content-based image retrieval. [Research Report] RR-4439, INRIA. 2002. ⟨inria-00072149⟩

Share

Metrics

Record views

172

Files downloads

365