ARHON: A multimedia database design for image documents

In this paper we present the design and initial implementation of a multimedia system for historical documents. The system, currently under development at FORTH, will be used by the Vikelea Municipal Library of archives at Heraklion, Crete. The material consists of historical documents that need to be digitized, indexed and continuously annotated by scholars so that the value of the archive will increase by its use. We present the design architecture and initial implementation of a client-server model for the creation, maintenance and use of such a digital archive. Also, we present a technique for marking-up different media to provide rapid entering of semantic information The user can interactively markup parts of the representation on the medium of choice (e.g. areas on a document image) and associate local information with text, images, sound-clips, video or sub-parts of them without altering the original. The output is XML files that can be parsed by viewers to navigate through the semantic information or used by search agents to refine the accuracy of retrieval in a query. For demonstration purposes we present ImageTagger, a prototype for document image tagging. Developing in Java allows for net-centric administration and use of a number of heterogeneous databases through standard Internet browsers.