Online Myanmar Handwritten Compound Words Recognition and Erratum Detection with MICR

Abstract

paper describes an effective recognition and detection erratum approach for Myanmar handwritten compound words. In this article, MICR (Myanmar Intelligent Character Recognition) method is used for the character recognition. The method is composed of statistical/semantic information and the final decision is made by the voting system. MICR has been successfully applied for many applications in online Myanmar character recognition field. The method recognizes isolated characters only and not compound words or phrases. Erratum Detection is a new technique, it detected irregularly form of isolated compound word in string texts. It dependent on the language set of string substitutions reflects the surface form of errors that result from cognitive, typographical mistakes, or mistyping. A robust erratum detection technique is needed to cover above all situation. The system index pair possible extended/medial code and then provides a pair code from the database for detect complete compound word. In detection, it has three situations: twice the same extended, extended/ medial pair not matching error, not compound word in real. Then the final output, Myanmar compound words will be produced editable text with highlight color in each error word.