Handwritten Arabic Character Recognition Based on Minimal Geometric Features

Abstract

On-line handwriting recognition is one of the most successful applications in the area of pattern recognition. Though this field is quite matured, yet the research issues are still challenging, particularly in handwriting character recognition, where the problems are still wide open. The OCR system for printed characters is almost done, though it cannot guarantee for 100% accuracy. However, the research works in recognition of Arabic handwriting are still at the beginning and require more attention. This paper presents the novel on-line Arabic handwriting character recognition. An efficient approach is introduced here to divide it into some particular component. A set of features are extracted from these components, and then encoded for the classification stage. The system classification is implemented by using two processes, i.e. weight initialization in back propagation, and with multilayer perceptron neural network. Finally, the proposed system was tested on a database of Arabic handwritten samples.