Segmentation of Bangla handwritten text

Abstract

Segmentation of handwritten text is the pre-process of recognition of handwritten text. Because of skewed lines, touching characters and different writing styles, segmentation of handwritten text is difficult than printed text. Correct segmentation is necessary to recognize the characters properly. That’s why a good segmentation is needed. Segmentation of Bangla handwritten text is a great challenge because of its complex characteristics. Till now there are very few works on Bangla handwritten text segmentation and still there are some problems that can’t be solved. After analyzing some methods and techniques that have already been done, here we propose a recursive algorithm for line, word and character segmentation. The algorithm works well if there is no overlapping situation between lines and characters as well.