A Similarity-Based Approach for Audiovisual Document Classification Using Temporal Relation Analysis

Zein Al Abidin Ibrahim,Isabelle Ferrane,Philippe Joly

doi:10.1155/2011/537372

Abstract

We propose a novel approach for video classification that bases on the analysis of the temporal relationships between the basic events in audiovisual documents. Starting from basic segmentation results, we define a new representation method that is called Temporal Relation Matrix (TRM). Each document is then described by a set of TRMs, the analysis of which makes events of a higher level stand out. This representation has been first designed to analyze any audiovisual document in order to find events that may well characterize its content and its structure. The aim of this work is to use this representation to compute a similarity measure between two documents. Approaches for audiovisual documents classification are presented and discussed. Experimentations are done on a set of 242 video documents and the results show the efficiency of our proposals.

Highlights

Motivated by the fact that large scale document indexing cannot be handled by human operators, researches tend to use high-level automatic indexing with the recent existing huge masses of digital data
We have proposed a novel approach for video classification that bases on the analysis of the temporal relationships between the basic events in audiovisual documents
We have based on some basic segmentation results in order to propose our new representation method called the Temporal Relation Matrix (TRM)

Summary

Introduction

Motivated by the fact that large scale document indexing cannot be handled by human operators, researches tend to use high-level automatic indexing with the recent existing huge masses of digital data. Finding the discriminating characteristics is still a challenging issue, especially if one wants to keep the detection of the basic events reliable and robust enough. Another challenge is to detect events of a sufficiently high semantic level and to produce indexes that are highly relevant according to the document content and structure. Such indexes will allow requests such as, “I am looking for an interview of Mister X by Miss Y about the movie Z” to be in a high-level information retrieval task

Objectives

Methods

Findings

Discussion

Conclusion