Multimodal Biometric Authentication System Based on Feature Level Fusion of Face, Fingerprint and Palm Print

Abstract

This paper presents the Multimodal biometric authentication system based on feature level fusion of multiple biometric traits like face, fingerprint and Palm print. The fusion can be of many types such as fusion at the Score level, Decision Level and at the Feature extraction level. The Fusion of biometric data at the score level and the decision level is easy to obtain. But the fusion of different biometric traits in feature extraction level is difficult to achieve. But the fusion at the feature extraction level will give better results than other fusion methods. Because the raw data obtained during the feature extraction level contain genuine data. The features are extracted by using the Scale Invariant Feature Transform (SIFT). SIFT generates Feature vectors. The Scale Invariant Feature Transform is the new approach in this paper for the fusion of multiple biometric traits. The method is to extract the individual feature vectors of face, fingerprint and palm print and then to concatenate them to form the fused features. These fused features are matched with the template values to authenticate a person.