Handwritten and Printed Word Identification Using Gray-scale Feature Vector and Decision Tree Classifier

Document image analysis is one of the important steps towards a paper free world. An effective Optical Character Recognition (OCR) system would be helpful for achieving this fit. But the next question may arise that whether a single OCR system will be sufficient for encoding both handwritten and printed text or not. So to come out of this dilemma, the work as reported here determines the category of a word from the document images containing words both in handwritten and printed forms. A 6- elements feature set is estimated from each gray level image and then these features are ranked based on discriminatory capabilities. Finally, a decision tree classifier has been designed and 1500 words images of handwritten and printed forms (equal in number) are fed to the classifier to evaluate the performance of the present technique. An overall success rate of 96.80% is achieved.