Erkennungssysteme und Dokumentanalyse: READ, Beitrag zum Abschlußbericht

Although electronic media is widely used within any organisation, paperwork is still a very important medium for information communication. The central goal of the READ project was the continuous improvement and design of new recognition systems, that would cover the gap between the electronic available information and the paper based one. The automatic processing of documents should be possible with the results of this project. In the research and development project READ a new method and tools for an adaptive handwriting recognition system were developed. Fuzzy clustering techniques supplement the fuzzy syntactic methods to extract the expert information automatically to create the required rule-base. By choosing a fuzzy paradigm, we were able to find an efficient solution for the classification of handwritten data. The restricted similarities of the existent proprietary document analysis systems were extracted. These similarities create the framework of the proposed design methodology for a generic document analysis system. The design methodology was supported by the development and test of a formalism to generalise the approach. The new design paradigm was tested successfully