Skip to Main content Skip to Navigation

Un modèle vectoriel relationnel de recherche d’information adapté aux images

Abstract : The increase of digital image acquisition devices, combined to the growth of the Web, requires the definition of Information Retrieval (IR) models and systems providing fast ac- cess to images searched by users among large amounts of data. The aim of our work is to define an IR model suited to images, integrating a rich semantics for representing these visual data as well as users’ queries, while allowing the application to large corpora. Our proposition merges the vector space model of IR – a widely tested model in textual IR – with Conceptual Graphs (CG) formalism, based on the use of star graphs (a star graph is an elementary CG made up of a single relation connected to some concepts representing image objects) as image descriptors. A novel weighting scheme (based on objects size and posi- tion, as on image heterogeneity) is defined and carefully validated in the experiments. We show that integrating relations into the vector space model through star graphs increases the system precision, and that the results are comparable with graph projection systems, while decreasing the processing time for users’ queries.
Complete list of metadatas

Cited literature [93 references]  Display  Hide  Download
Contributor : Jean Martinet <>
Submitted on : Friday, January 26, 2018 - 11:29:50 AM
Last modification on : Saturday, January 27, 2018 - 1:05:29 AM
Long-term archiving on: : Friday, May 25, 2018 - 2:13:32 AM


Files produced by the author(s)


  • HAL Id : tel-01693438, version 1



Jean Martinet. Un modèle vectoriel relationnel de recherche d’information adapté aux images. Vision par ordinateur et reconnaissance de formes [cs.CV]. Université Joseph Fourier (Grenoble I), 2004. Français. ⟨tel-01693438⟩



Record views


Files downloads