Title: An approach to offline handwritten Devanagari word segmentation

Authors: O.V. Ramana Murthy; M. Hanmandlu

Addresses: Department of Electrical Engineering, IIT Delhi, India. ' Department of Electrical Engineering, IIT Delhi, India

Abstract: Hindi is very popular language after Mandarin and English. Its script is Devanagari. Although major work is reported on OCR techniques for machine printed Devanagari script, very few works are beginning to report on offline handwritten Devanagari OCR. Particularly, very few works have been reported so far on segmentation of Devanagari words. This paper paves a step in that direction and open avenues for researchers in that field. A novel segmentation approach is proposed for segmentation of offline handwritten Devanagari words using techniques of Hough transform and connected components. The difficulties for segmentation in Devanagari script and systematic steps to accommodate those difficulties as much as possible have been presented with elaborate results.

Keywords: Devanagari word segmentation; Hough transform; corner detection; connected components; handwriting; offline handwritten words; Hindi; OCR; optical character recognition.

DOI: 10.1504/IJCAT.2012.050116

International Journal of Computer Applications in Technology, 2012 Vol.44 No.4, pp.284 - 292

Published online: 29 Oct 2012 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article