Handwritten script recognition system for languages with diacritic signs

Abstract

Automatic recognition of natural writing with totally unconstrained words remains a very challenging problem in artificial intelligence and pattern recognition. It is due to the large variety of handwriting styles, ambiguous boundaries of character strings and considerable variation in letter form. In this paper an off-line handwritten recognition system based on analytical approach, neural classification of characters, fuzzy rules and postprocessing by lexical analysis is presented. We afforded a special attention to the segmentation phase, as well as to the treatment of letters having diacritic signs, like in Romanian or other languages.