I.J. Modern Education and Computer Science, 2015, 5, 8-15 Published Online May 2015 in MECS (http://www.mecs-press.org/) DOI: 10.5815/ijmecs.2015.05.02 Copyright © 2015 MECS I.J. Modern Education and Computer Science, 2015, 5, 8-15 Person Authentication using Relevance Vector Machine (RVM) for Face and Fingerprint Long B. Tran Computer Science Department, University of Lac Hong, DongNai, 71000, VietNam Email: tblong@lhu.edu.vn Thai H. Le Computer Science Department, VNUHCM - University of Science, 70000, VietNam Email: lhthai@fit.hcmus.edu.vn Abstract—Multimodal biometric systems have proven more efficient in personal verification or identification than single biometric ones, so it is also a focus of this paper. Particularly, in the paper, the authors present a multimodal biometric system in which features from face and fingerprint images are extracted using Zernike Moment (ZM), the personal authentication is done using Relevance Vector Machine (RVM) and feature-level fusion technique. The proposed system has proven its remarkable ability to overcome the limitations of uni- modal biometric systems and to tolerate local variations in the face or fingerprint image of an individual. Also, the achieved experimental results have demonstrated that using RVM can assure a higher level of forge resistance and enables faster authentication than the state-of-the-art technique , namely the support vector machine (SVM). Index Terms—Multimodal Biometric; Feature Level Fusion; Face; Fingerprint; Recognition System; Relevance vector machine; Zernike moment I. INTRODUCTION Biometric-based personal authentication systems are estimated more reliable than traditional systems as their performance bases on a person‘s physiological and behavioral traits which are always authentic and secured[1], [2], [3] while traditional systems use passwords, key, ID cards which can easily be lost, shared, stolen or even forgotten [4]. Currently available biometric systems use a variety of physical or behavioral characteristics such as fingerprint, facial thermo-grams, face, hand/finger geometry, iris, retina, gait, signature, voice pattern and hand vein to establish identity [3],[5] (Figure.1). Each Biometric trait has its own advantages and disadvantages; however, a biometric system would be considered admissible when it has these characteristics: universality, uniqueness, permanence, measurability, performance, acceptability and circumvention. [6]. Although uni-biometric systems, those using single biometric traits, are currently popular in use for their recent significant progress, they still suffer some drawbacks that impede their efficiency, namely noisy data, restricted degree of freedom, intra-class variability, non-universality, spoof attack and unacceptable error rates. However, with the feature of utilizing different biometric traits, such as different sensors, multiple samples of the same biometrics, different feature representations, or multi-modalities, multi-biometric systems can alleviate many of the limitations faced by uni-biometric system [7]. Fig. 1. Examples of biometric characteristic Multimodal biometric systems have gained intensive interest among designers and practitioners for two reasons. First, these systems perform better than uni- modal ones; second, their speed is improved satisfactorily. This leads us to the hypothesis that with the employment of multiple modalities (face and fingerprint), our proposed system would avoid the mentioned drawbacks of modality- based techniques. According to Ross and Jain [7], in a multimodal biometric system, fusion can be performed at different levels, such as sensor, feature, matching score and decision. Among which [8] feature level fusion is usually considered difficult as different biometrics would have different feature vectors and different measures. Support Vector Machine (SVM) classifiers have proven in the literature its outstanding performance of authentication tasks [9][10][11]. However, SVM requires sufficient amount of training data and a high number of support vectors for its performance, leading to expensive fusion. Regarding this, in this paper an authentication approach using Relevance Vector Machine (RVM) [12] is Person Finger print Face Iris Palmprint Hand vein