Leaders: Nozha Boujemaa and Alexis Joly INRIA-IMEDIA
Partners:

• INRIA - IMEDIA team (France), A. Joly, N. Boujemaa, J. Law-To
• INA (Institut National de l'Audiovisuel, France), D.Terouggi, O. Buisson, J. Law-to (joint with INRIA)
• INRIA -- VISTA team (France), I. Laptev

Objectives: Due to the increasing broadcasting of digital content (TV Channels, Web, Video Blogs, etc), finding copies in a large video and photo database has become a critical new issue and Content Based Copy Detection presents an alternative to the watermarking approach to identify video sequences and still images. It provides a useful tool to control the copyrights of digital contents on the web and on TV, and also a powerful tool to analyze and explore these data. A crucial difficulty is the fundamental difference between a copy and the notion of similar image encountered in Content-Based Retrieval: a copy is not an identical or a near replicated video sequence but rather a transformed video sequence. These photometric or geometric transformations (gamma and contrast transformations, overlay, shift, etc) can greatly modify the signal, and therefore a copy can in fact be visually less similar than other kinds of videos that might be considered similar. The goal of this showcase is to introduce the motivations and the issues of copy detection on the web and in TV streams and to present efficient methods for video and still image.

Activities: We developed a demonstrator to be presented in professional meetings and scientific conferences. This demonstrator can be used either to browse some precomputed interesting monitoring results or for an interactive demo where the user can retrieve a video "on the fly". A dedicated interface was developed with the following functionalities:

• a double player for the candidate and the retrieved video with synchro or independent playing options. Resynchronisation button
• sound controls (both sounds, just candidate, just retrieved, none)
• thumbnails list representation of all results
• Tree view of results

 
A first version of the demonstrator was presented at the CeBit professional meeting (4-9 march 2007 in Hannover)

INRIA VISTA team is currently developping a new method for content-based copy detection based on spatio-temporal interest points. This technique will be integrated in IMEDIA or INA retrieval system in order to show the improvements this technique can bring.

IMEDIA developped last year a technique dedicated to still images that will be also integrated in the demonstrator.

The final complete demonstrator will be presented in CIVR 2007 and in IBC 2007.