Abstract—Character recognition plays an important role in the modern world. It can solve more complex problems and make human's job easier. An example is handwritten numerals recognition. This system can be used for many applications such as automatic recognition of the courtesy amount in bank cheques and recognition of postal code, as they require a high accuracy of recognition, and include many variations.
In this paper, we have presented a new algorithm for segmenting connected Eastern Arabic numerals, a hybrid classifier to improve recognition accuracy. The hybrid classifier consists of Multilayer neural network and decision tree. The testing of the proposed recognition system indicates that the classification accuracy is in average 83%.

Index Terms—Character Recognition, Segmentation, Neural Networks, Classification.

Marco Alfonse, Computer Science Department, Faculty of Computer & Information Sciences, Ain Shams University (e-mail: marco_alfonse@yahoo.com).
Mohamed Almorsy, Computer Science Department, Faculty of Computer & Information Sciences, Ain Shams University (e-mail: mohamed_almorsy@hotmail.com).
M. S. Barakat, Computer Science Department, Faculty of Computer & Information Sciences, Ain Shams University (e-mail: msamirhb@hotmail.com).

Cite: Marco Alfonse, Mohamed Almorsy, and Mohamed Samir Barakat, "Eastern Arabic Handwritten Numerals Recognition," International Journal of Computer and Electrical Engineering vol. 2, no. 2, pp. 277-282, 2010.