VideoText database systems

Abstract

This paper introduces a new approach to realize video databases. It consists of a VideoText data model based on free text annotations associated with logical video segments and a corresponding query language. Traditional database techniques are inadequate for exploiting queries on unstructured data like video, supporting temporal queries, and ranking query results according to their relevance to the query. We propose to integrate information retrieval techniques to provide these features and to extend the query language to accommodate interval queries that are useful due to the stream nature of video data. Algorithms are presented to show how user queries are evaluated. Finally, a generic and modular video database architecture based on the VideoText data model is described.