Journals Proceedings

Abstract

Arabic script character recognition remains a challenging task due to its cursive nature. The aim of feature extraction from the input strokes reduces the input pattern to avoid complexities while maintaining the high accuracy. Feature extraction in pattern recognition problems involves the extraction of unique and salient patterns from the preprocessed data in order to enhance the discriminatory power and reduce the data for classification and it is crucial for the success of recognition system. We proposed feature level fusion for Arabic character recognition. We extracted structural and directional features for handwritten stroke and fused these feature to form more discriminant feature matrix. The feature level fusion for handwritten character recognition provides considerable gain in accuracy. The fusion of feature is also suitable for other handwritten tasks i.e. personality determination, writer identification where the more variations are required.