AN INTERFACE FOR IMAGE RETRIEVAL AND ITS EXTENSION TO VIDEO RETRIEVAL - Archive ouverte HAL Access content directly
Conference Papers Year : 2006

AN INTERFACE FOR IMAGE RETRIEVAL AND ITS EXTENSION TO VIDEO RETRIEVAL

(1) , (2) , (1)
1
2

Abstract

Semantic video retrieval is still an open problem. While many works exist in analyzing the video contents, few ones present the retrieval results to the users and interact with him/her. In this article, firstly, we propose a 2D graphic interface adapted to the problem of image retrieval that enables a bidirectional communication: from the system towards the user to visualize the current research results and from the user towards the system so that the user can provide some relevance feedback information to refine his/her query. In this interface, the visualization shows the image query in the middle of the screen and the result images in a 2D plan with distances showing the similarity measures between images and the query. We propose also a method of relevance feedback in form of validation, in this interface, for image retrieval. This approach has been implemented and tested with different image databases. Secondly, we analyze the extension of this approach for video retrieval. For this, we extract the key frames from video and use them to represent the research results of video as well as to do the relevance feedback.
Fichier principal
Vignette du fichier
llethi_aboucher_mthonnat.pdf (491.04 Ko) Télécharger le fichier
Origin : Files produced by the author(s)
Loading...

Dates and versions

inria-00186400 , version 1 (09-11-2007)

Identifiers

  • HAL Id : inria-00186400 , version 1

Cite

Thi Lan Le, Alain Boucher, Monique Thonnat. AN INTERFACE FOR IMAGE RETRIEVAL AND ITS EXTENSION TO VIDEO RETRIEVAL. Proceedings of the Third National Symposium on Research, Development and Application of Information and Communication Technology (ICT. rda'06), May 2006, Hanoi, Vietnam. ⟨inria-00186400⟩

Collections

INRIA INRIA2
263 View
217 Download

Share

Gmail Facebook Twitter LinkedIn More