Egyptian Computer Science Journal Vol. 38 No.3 September 2014 ISSN-1110-2586 -75- A Palmprint-Based Identification System using Radon Transform Shimaa M. Al-seddek, Hassan H. Soliman, Muhammad S. Morsy , Sherif S. Kishk ECE Department, Faculty of Engineering, Mansoura University, Mansoura – Egypt sh_sediek@yahoo.com, hhsoliman@yahoo.com, drmmorsy@yahoo.com, shkishk@mans.edu.eg Abstract This paper proposes a palmprint-based identification system using Radon transform. Palmprint is a reliable identification method because the print patterns are unique even in the monozygotic twins, has permanent ridge structure, has fixed line structure, user friendliness, low cost capturing devices, and low resolution imaging. The proposed system is applied to CASIA database. Radon transform is used for extracting the features because it can be used in palm lines detection with high precision and efficiency. Live and enrolled palmprint are matched using Euclidean distance for verification. This system can achieve Equal Error Rate (EER) of 0.47% and Genuine Acceptance Rate (GAR) of 99.56%. The results of this study showed that the proposed system achieves higher verification accuracy than other palmprint identification systems. Keywords: Identification, Biometrics, CASIA, Palmprint, Radon Transform, Euclidean Distance. 1. Introduction Personal identification systems using biometric features have been widely used in critical security applications such as banking, access control and crime investigation. Examples of these biometric features include Fingerprint, Iris, Face, Hand geometry, Palmprint, etc. Fingerprint and Face recognition are the most widely studied biometrics. The reliability of Face biometrics has been hampered by the problems caused by pose, expressions and illumination. For decades now, Fingerprint is the most effective identification system. However, it also has some limitations as different groups of users such as elderly people and manual workers fail to deliver good quality fingerprint images [1]. In addition, noise caused by multiple fingerprints on the device sensor negatively affects the efficiency of the identification system. Iris recognition is another reliable method, but its acquisition device is relatively expensive and it is not very convenient to collect [2]. Of all the above-mentioned systems, the Palmprint-based identification system has been intensively developed because of its advantages over other features. Compared to other biometric systems, palmprint has many advantages such as high accuracy, user friendliness and low resolution imaging that can be employed in palmprint recognition that is based on creases and palm lines, making it possible to perform well in real time pre-processing and feature extraction steps. It has a large area for extracting features and is more resistant to injuries and dirt compared to fingerprint. In these cases, the distinguishable features rely on palm lines and texture patterns [3]. These specific line features are extracted after applying effective pre-processing filters to enhance the Region of Interest (ROI) of palmprint images.