An Efficient Character Segmentation Approach for Handwritten Hindi Text

OCR remains a challenging area of research for past few years. Accuracy of OCR primarily depends on how correctly characters are segmented. This makes segmentation a crucial step. This work describes the novel technique for character segmentation of handwritten Hindi text using basic structural properties. The proposed method first uses the polygonal approximation to get the digitally straight line segment of the word. On these segments, we apply traversing algorithm which results into segmentation of characters. The success rate is 95.70%, which is satisfactory comparing with other existing methods.