Optical Character Recognition (OCR) attempts to convert large amounts of documents into electronic form for further processing. The most important stage of an OCR system is the feature extraction stage, because the recognition performance depends on the features selected at this stage. In this presentation, the state-of-the-art in feature extraction methods for isolated handwritten characters and unconstrained words is discussed, as well as a novel feature extraction is presented. Several databases either contemporary or historical are used in order to demonstrate the efficiency of the proposed methodology.
Talk slides in ppt [~2,9MB]http://www.iit.demokritos.gr/docs/seminars/Off-Line_Handwritten_OCR.ppt