Graphical tools and techniques for querying document image databases

Abstract

This paper describes document models and relations for the retrieval of document images. The underlying methodology was developed for the Intelligent Document Image Retrieval System (IDIR), which aims to extend document image database query capabilities. Traditional component type and keyword features are insufficient in describing logical and structural aspects of documents nature.

We have developed the necessary object-oriented document models to carry out complex multi domain retrieval scenarios. In this paper we focus on retrieval capabilities and underlying methodology that supports different schemes. For these models and query schemes (QS), new graphical techniques are introduced. The IDIR allows complex combinations of different QS's, using the extended concept of `frame logic', developed in our earlier work for the attribute management. Furthermore, a concept of document similarity is introduced with relations to QS's, document models, structure and role of use. Examples are shown for the retrieval of document images from University of Washington Database.