Semantic video content indexing and retrieval using conceptual graphs

Abstract

This paper proposes a model for the indexing and retrieval of semantic video content using conceptual graphs. Conceptual graph based indexing is even better since not only nonambiguous concepts are used but also relations between these concepts are indexed. The proposed model allows video content representation both at the semantic and at the signal levels. The signal based representation is done via the perceptive view using classical signal descriptors. The representation by conceptual graph allows a more precise description of the content and it also allows the linking between the involved concepts and the sub Medias of the video documents. The various concepts and relations involved in the model can be taken from general and/or domain specific ontologies and completed by lists of instances. the correspondence between the query and any document is searched as a projection of the graph representing the query on the graph representing the document. This projection is done using the substitution rules permitted by the ontology.