Research into the theory and practice of visual information retrieval is of long standing at the University of Brighton and is led by Professor Peter Enser.
This work is set in the context of a visually oriented society which has harnessed successive technologies to facilitate the creation and transmission of records in the visual medium, without a corresponding facility for gaining effective access to the knowledge encoded within those records.
Various leading image archives and libraries have collaborated in a number of externally funded studies undertaken at Brighton, the main focus of enquiry being the analysis of user need for both still and moving images. These studies have been designed to provide an informed view on the functionality of image indexing and retrieval strategies, an issue which has come to prominence in the research agenda for information retrieval and computer science in recent years owing to new, digitally-enabled capabilities in image processing and management.
Of particular significance are those capabilities derived from Content-Based Image Retrieval (CBIR) techniques which enable images to be retrieved on the basis of quantifiable attributes - colour, texture or shape - derived from their pixel structure. The need to bridge the 'semantic gap' between such low level CBIR processes and the high level semantics which our research has shown to typify user needs in the visual realm is a primary focus for our current research activity. Ongoing effort is focused on metadata construction techniques for hybrid text/CBIR models, the contribution which evolving ideas in taxonomies and ontologies might make to these construction techniques, and usability analysis associated with the migration from a human-mediated to a computer-mediated image retrieval environment.
Our current research is funded by the Arts and Humanities Research Council, prior to this the main sources of funding for our VIR work have been the antecedents of the Council for Museums, Archives & Libraries, and the work has been disseminated within the library and information science, cultural heritage and computer science communities. A complete list of publications relating to our visual information retrieval research is shown on the VIR Publications pages.
