Document and handwriting analysis for archive index card conversion

Museums and libraries have an urgent need to find ways of computerising archive card indexes to support computational analysis and automated database interrogation of their curatorial records. Existing commercial OCR software and document analysis packages are totally inadequate for handling the diverse heterogeneous formats of typescript, hand-print and handwriting encountered in archive indexes. This paper outlines the problems to be overcome in transcribing archive indexes, and proposes a progressive approach to archive conversion in which a hierarchical interpretation of each card image is recorded for posterity. Systems developed using the tools proposed will have application in museums, libraries and archives all over the world, establishing this field as a major future commercial application of document and handwriting analysis