---

Nowadays, online handwritten signature verification has become an extremely active area of research. Significant results have been achieved in terms accuracy and computation so far. However, it is evident that there is still a room for improvement either in accuracy or computation speed. This paper proposes online handwritten signature verification system. The system allows a user to register and verify the signature. In addition, the approach of using the combination of local and global features is presented. More importantly, the approach of using a dynamic threshold along with fixed thresholds is also presented. In this study, we used five signatures as references, which was found to be very promising. Lastly, we evaluated the system performance by creating a local database consisting of 20 participants (users) with two sessions. The overall performance of the system is encouraging since the proposed system yields 2% of EER, which is considered as a reasonable performance level for most environments.

---

Rebwar M. Nabi finished his MSc in Advanced Computer Science at Newcastle University in the United Kingdom in 2012. Currently, he is a Ph.D. student at Sulaimani Polytechnic University in Machine Learning. He is the co-founder of Kurdistan Technical Institute and Qaiwan University - UTM Franchise.

---