A system is proposed to recognize handwritten numerals in both Devnagri (Hindi) and English. It is assumed at a time the numerals will be of one of the above two scripts and there are no mixed script numerals in an input string. A set of global and local features, which are derived from the right and left projection profiles of the numeral image, are used. During experiments it was found that the Devnagri numeral set had a much better recognition and rejection rate as compared to the English character set and so the input numeral is first tested by the Devnagri module. The correct recognition enables the system to be set to the appropriate context (Devnagri/English numeral set). Subsequent identification of the other numerals is carried out in that context only.
G. S. Lehal, Nivedan Bhatt