Artificial immune system for handwritten Arabic word recognition

In this paper, a handwritten Arabic word recognition system is proposed. Specifically, the use of Artificial immune recognition system to recognize Arabic script in a medium vocabulary, is investigated. Relevant features are obtained using the Ridgelet transform. Experiments are conducted on samples extracted from the well-known IFN/ENIT dataset which contains names of Tunisian towns. The performance evaluation is carried out comparatively to SVM classifiers. Recognition rate highlight the efficiency of the AIRS classifier.