부산대학교 도서관

The considerable development of electronic document management system in these last years has led to a large increase in the number of documents must be controlled and archived, etc. Therefore, this is important in forensic investigations, because identifying the writer could assist in solving a crime. However, there is a lack of works done in the case of online Arabic writer identification. In this paper, we propose a novel system to text independent writer identification from online Arabic handwriting. Our proposed system is based on the use of Beta-elliptic model in feature extraction. Moreover, we explore the potential utility of Deep Bidirectional Long Short-Term Memory in classification. Experimental results admitted that the proposed system has a very good performance compared to the existing online Arabic writer identification systems.