Semantic Description of Multimedia Documents: the MPEG-7 Approach

This article addresses a possible approach on how to index a multimedia document in a semantic manner. At present, both automatic and semi-automatic tools allow to index a document using low level features, obtaining structural information like shot segmentation, speaker separation and so on. In spite of its utility, this information is too elementary to perform some tasks like querying, retrieval or authoring of other descriptions. Instead enabling powerful description schemes that allow to describe things like semantic time and place, concepts, relations among them and others, ensures an effective description of the semantic content of documents. In particular we analyze a description scheme, the Semantic DS, developed within the MPEG-7 standardization process, and we give an explanation of its components and examples for their usage.