Indexing and mining audiovisual data

  • Authors:
  • Pierre Morizet-Mahoudeaux;Bruno Bachimont

  • Affiliations:
  • Department of Computer Science and Engineering, UMR-CNRS 6599 – Heudiasyc, University of Technology of Compiègne, France;Department of Computer Science and Engineering, UMR-CNRS 6599 – Heudiasyc, University of Technology of Compiègne, France

  • Venue:
  • AM'03 Proceedings of the Second international conference on Active Mining
  • Year:
  • 2003

Quantified Score

Hi-index 0.00

Visualization

Abstract

In this paper we present a review of recent research and development works, which have been developed in the domain of indexing and mining audio-visual document. We first present the characteristics of the audio-visual documents and the outcomes of digitising this kind of documents. It raises several important issues concerning the new definition of what is a document, what is indexing and what are the numeric principles and technologies available for performing indexing and mining tasks. The analysis of these issue let us introduce the notion of temporal and multimedia objects, and the presentation of the three steps for indexing multimedia documents. It includes the clear distinction between descriptors and indexing. Finally we introduce the MPEG-7 paradigm, which sets the technical environment for developing indexing applications; Then we shortly review current developments, based on the text mining, the XML-Schema, and the event description interface approaches.