When images work faster than words: The integration of content-based image retrieval with the Northumbria Watermark Archive

Information on the manufacture, history, provenance, identification, care and conservation of paper-based artwork/objects is disparate and not always readily available. The Northumbria Watermark Archive will incorporate such material into a database, which will be made freely available on the Internet providing an invaluable resource for conservation, research and education. The efficiency of a database is highly dependant on its search mechanism. Text based mechanisms are frequently ineffective when a range of descriptive terminologies might be used i.e. when describing images or translating from foreign languages. In such cases a Content Based Image Retrieval (CBIR) system can be more effective. Watermarks provide paper with unique visual identification characteristics and have been used to provide a point of entry to the archive that is more efficient and effective than a text based search mechanism. The research carried out has the potential to be applied to any numerically large collection of images with distinctive features of colour, shape or texture i.e. coins, architectural features, picture frame profiles, hallmarks, Japanese artists stamps etc. Although the establishment of an electronic archive incorporating a CBIR system can undoubtedly improve access to large collections of images and related data, the development is rarely trouble free. This paper discusses some of the issues that must be considered i.e. collaboration between disciplines; project management; copying and digitising objects; content based image retrieval; the Northumbria Watermark Archive; the use of standardised terminology within a database as well as copyright issues.