Recognition of Farsi Handwritten Numbers Using the Fuzzy Method

There are wide varieties of handwritten characters which differ not only from person to person butalso from the state of mood of the same person. Nevertheless humans are trained to extract thespecific features characterizing a symbol. This paper aims to introduce fuzziness in the definitionof the proposed pattern features, which provides the enhancement to the handwritten characterinformation to be stored. Some novel shape features in fuzzy linguistic domain are proposed. Theexperimental results indicate 95 % accuracy on recognition of Farsi numbers over the selecteddatabase.