OIAHCR: Online Isolated Arabic Handwritten Character Recognition Using Neural Network

Basem Al-Ijla and Kathrein Kwaik
      Faculty of Information Technology, Islamic University Of Gaza, Palestine

 

Abstract: In this paper, an online Isolated Arabic Handwritten Character Recognition system is introduced. The system can be adapted to achieve the demands of hand-held and digital tablet applications. To achieve this goal, despite of single Neural Networks, four Neural Networks are used, one for each cluster of characters. Feed Forward Back Propagation Neural Networks are used in classification process. This approach is employed as classifiers due to the low computation overhead during training and recall process. The system recognizes On-Line isolated Arabic character and achieves an accuracy rate 95.7% from untrained writers and 99.1% for trained writers.

Keywords: Back propagation, classification, feature extraction, feature selection, feed forward neural networks, and Optical Character Recognition (OCR).

  

Full Text