The International Arab Journal of Information Technology (IAJIT)


Arabic Text Recognition

The issue of handwritten character recognition is still a big challenge to the scientific community. Several approaches to address this challenge have been attempted in the last years, mostly focusing on the English pre-printed or handwritten characters space. Thus, the need to attempt a research related to Arabic handwritten text recognition. Algorithms based on neural networks have proved to give better results than conventional methods when applied to problems where the decision rules of the classification problem are not clearly defined. Two neural networks were built to classify already segmented characters of handwritten Arabic text. The two neural networks correctly recognized 73% of the characters. However, one hurdle was encountered in the above scenario, which can be summarized as follows: there are a lot of handwritten characters that can be segmented and classified into two or more different classes depending on whether they are looked at separately, or in a word, or even in a sentence. In other words, character classification, especially handwritten Arabic characters, depends largely on contextual information, not only on topographic features extracted from these characters.


[1] Al-Badr A. and Mahmoud A., “Survey and Bibliography of Arabic Optical Text Recognition,” Signal Processing, vol. 41, pp. 49- 77, 1995.

[2] Al-Yousefi H. and Upda A., “Recognition of Arabic Characters,” IEEE Pami, vol. 14, no. 8, pp. 853-857, 1992.

[3] Almeida L., Multilayer Perceptrons - Handbook of Neural Computation, IOP Publishing Ltd and Oxford University Press, vol. 30, 1997.

[4] Bell J. and Zemanec P., “Test of Two Arabic OCR Programs,” Distributed on Reader 14 January and Itisalat 17 January, 1995.

[5] Burges C., Ben J., Denker J., Lecun Y., and Nohl C., “Off Line Recognition of Handwritten Postal Words Using Neural Networks,” International Journal of Pattern Recognition and Artificial Intelligence, pp. 689-704, 1993.

[6] Grother P. and Loeve K., “Feature Extraction for Neural Handwritten Character Recognition,” in Proceedings of SPIE, pp. 155-166, 1992.

[7] Han K. and Sethi I., “Off-Line Cursive Handwriting Segmentation,” in Proceedings of ICDAR’95, Montreal, Canada, pp. 894-897, 1995. Arabic Text Recognition 163

[8] Haraty R. and Hamid A., “Segmenting Handwritten Arabic Text,” ACIS International Journal of Computer and Information Science (IJCIS), vol. 3, no. 4, December 2002.

[9] Haraty R. and Zabadani H., “ABJAD: An Off- Line Arabic Handwritten Recognition System,” in Proceedings of the 2002 International Arab Conference on Information Technology (ACIT'2002), Doha, Qatar, December 2002.

[10] Khorsheed M. and Clocksin W., “Structural Features of Cursive Arabic Script,” (BMVC'99), pp. 422-431, 1999.

[11] LeCun Y., Boser B., Denke J., and Jackel L., “Back-Propagation Applied to Handwritten Zip Code Recognition,” Neural Computation, pp. 541-551, 1989.

[12] Sakhr Software,, 2003. Ramzi Haraty is an assistant professor of computer science at the Lebanese American University in Beirut, Lebanon. He received his BSc and MSc degrees in computer science from Minnesota State University-Mankato, Minnesota, and his PhD in computer science from North Dakota State University-Fargo, North Dakota. His research interests include database management systems, artificial intelligence, and multilevel secure systems engineering. He has well over 50 journal and conference paper publications. He is a member of Association of Computing Machinery, Arab Computer Society and International Society for Computers and their Applications. Catherine Ghaddar received her MSc degree in computer science from the Lebanese American University in Beirut, Lebanon. Her research interests include database management systems, neural networks, and software engineering.