The segmentation of unconstrained handwritten text lines into words is an important stage in word recognition systems. This paper addresses a methodology to overcome the challenges, which are amplified by the non-uniform spaces between words and overlapping components by using a few statistical approaches. The system was developed using Java 2 and ImageJ tool. In this approach, a text line image is scanned vertically, holding only the spatial information. A scheme based on distance metrics and gap classification into inter-word gap and intra-word gap is presented. The threshold value is determined by using arithmetic mean, inter-quartile mean or trimmed mean based on the variation in the text. A pre-processing of removal of noise and correction of skew angle and dominant slant angle were done to improve the recognition accuracy. The system was illustrated with a few cases. A quantitative analysis of the experiment done on the system by using 1100 text lines from IAM database achieved an accuracy of 96. 72% and found the system faster and reliable. Further, the proposed method is compared with the contour based and non-contour based techniques.

1. Marti. U. V. and Bunke, H. 2001. Text line segmentation and word recognition in a system for general writer independent handwriting recognition. In: Proceedings of International Conference on Document Analysis and Recognition, 159-163.
2. Marti, U. V. and Bunke, H. 2002. The IAM-Database: an English sentence database for offline handwriting recognition. International Journal on Document Analysis and Recognition, 5, 39-46.
3. Seni, G. and Cohen, E. 1994. External word segmentation of offline handwritten text lines. Pattern Recognition, 41-52.
4. Manmatha, R. and Rothfeder, J. L. 2005. A scale space approach for automatically segmenting words from historical handwritten documents. In: IEEE Transactions on Pattern Analysis and Machine Intelligence, 27, 1212-1225.
5. Papavassiliou, V. , Stafylakis, T. , Katsouros, V. , and Carayannis, G. : Handwritten document image segmentation into text lines and words. Pattern Recognition, 43, 369-377.
6. Simistira, F. , Papavassiliou, V. , Stafylakis, T. , Katsouros, and V. , Carayannis, G. (2011). Enhancing handwritten word segmentation by employing local spatial features. In: Proceedings of International Conference on Document Analysis and Recognition, 1314-1318.
7. Louloudis, G. , Gatos, B. , Pratikakis, I. and Halatsis, C. : Line and word segmentation of handwritten documents. In 1st International Conference on Frontiers in Handwriting Recognition (ICFHR), 247-252.
8. Louloudis, G. , Gatos, B. , Pratikakis, I. , and Halatsis, C. : Text line and word segmentation of handwritten documents. Pattern Recognition Journal. Special issue on Handwriting Recognition.
9. Louloudis, G. , Stamatopoulos, N. , and Gatos, B. 2009. A Novel two stage evaluation methodology for word segmentation technique. In: Proceedings of International Conference on Document Analysis and Recognition, 686-690.
10. Lemaitre, A. , Camillerapp, J. , and Couasnon, B. 2011. A perceptive method for handwritten text segmentation. Document Recognition and Retrieval, XVIII.
11. Kuniawan, F. , Khan, A. R. , and Mohamad, D. 2009. Contour vs Non-Contour based word segmentation from handwritten textlines: an experimental analysis. In: International Journal of Digital Content Technology and its Applications, 3(2).
12. Otsu, N. 1979. A Threshold selection method from gray level histograms. IEEE Transaction on Systems, Man and Cybernetics, 9(1), 62-66.

Inter-quartile Mean Projection Profile Connected Component Distance Metrics