Sample Sidebar Module

This is a sample module published to the sidebar_top position, using the -sidebar module class suffix. There is also a sidebar_bottom position below the menu.

Sample Sidebar Module

This is a sample module published to the sidebar_bottom position, using the -sidebar module class suffix. There is also a sidebar_top position below the search.
قسم الحاسبات

1HadeelJ.Jriash *      &       2Dr. Nada A. Z. Abdullah

1,2College of Science, Computer Science Department, Baghdad University, Baghdad, Iraq

Emails:1hadeeljabbir@yahoo.com    , 2nadaazah@scbaghdad.edu.iq 

Abstract:

Signature of a person is an important biometric attribute of a human being which can be used to authenticate human identity. Various complex methodologies in the past have been proposed for signature verification through feature extraction. This paper presents new method for handwritten signatures verification system to determine if the signature is original or not, that depends on a Discrete Radon Transform (DRT) as feature extraction, Euclidian Distance (ED) and Probabilistic Neural Network (PNN) are usedas verifiers. Different sets of angles are used (4, 6, 8,and  10) in the DRT. The feature vector is reduced using Mean_Max method. The system trained using handwritten English dataset for 55 signers. Satisfactory results are obtained with 90.3% accuracy of system when using the ED,the PNN has the average accuracy of 90.8 % when using single net for each signer and the PNN has the average accuracy of 95.5 %when using single net for all signers. 5.1% Equal Error Rate (EER), 5.8% False Acceptance Rate (FAR), 4.5% False Rejection Rate (FAR) for skilled forgeries on our independent database.These rates are considered high when compared with the results of related works.