The objective of this research was to improve an automatic writer recognition technique for off-line Arabic
handwritten text. This objective was met by developing a new method which not only addressed the
problems in existing techniques but also realized better outcomes than the current solutions on this issue.
The projected method is based on Discrete Wavelet Transform (DWT) to extract feature. Also important
features for Arabic handwritten alphanumeric character recognition to aid the verification step. This
scheme involves of four phases: preprocessing, feature extraction, important features and finally writing
recognition. Important features are constructed by data reduction method. After collecting all important
features of each character. To recognize fonts, which improves the classification performance. Then
decision tree classifier was used for writing recognition. In this paper provide good accuracy dealing with
important attributes. The suggested ideal is fast and reliable. The proposed technique has been achieved
very promising results, with a validation accuracy of 98.45% for HACDB database.
الاستاذ المساعد الدكتورة اسماء شاكر عاشور | Assist. Prof. Dr. Asmaa Shaker Ashoor | 18402