A Technique for Classification of Printed and Handwritten Text

Machine printed and handwritten words are sometimes mixed in a single document like in data entry forms. Since the algorithms for recognition of machine-printed and handwritten text are based on different techniques, so it is necessary to separate between these two types of texts before feeding it to respective optical character recognition systems. This separation will definitely increase the performance and overall system quality. Handwritten/machine-printed classification is the process to discriminate handwritten from machine-printed text and is a challenging task. It includes two issues first is detecting the letters and then classifier will classify the machine written text and hand written text. The proposed techniques is based on structural features of text i.e. aspect ratio of the text is calculated to discriminate between handwritten and machine printed text and results show the effectiveness of proposed approach.