Abstract
Machine printed and handwritten texts intermixed appear in the ICR cells of variety of documents. Recognition techniques for machine printed and handwritten text in these document images are significantly different. It is necessary to separate these two types of texts and feed them to the respective engine - OCR (Optical Character Recognition) and ICR (Intelligent Character Recognition) engine to achieve optimal performance. This paper addresses the problem of classification of machine printed and handwritten text from acquired document images. Document processors can increase their productivity and classify handwritten and printed characters inside the ICR cells and feed their images to the appropriate OCR or ICR engine for better accuracy. The algorithm is tested on variety of forms and the recognition rate is calculated to be over 91%.