Multi-modal biometric authentication fusing iris and palmprint based on GMM

Biometrics is an effective technology for personnel identity authentication (PIA), but unimodal biometric systems which use a single trait for authentication, will suffer from problems like noisy sensor data, nonuniversality, lack of distinctiveness of the biometric trait, unacceptable error rates, and spoof attacks. These problems can be tackled by using multi-biometrics in the system. This paper investigates the fusion of palmprint and iris biometric features. A new fusion scheme at score level that combines Gaussian mixture model (GMM) and score normalization is proposed. The features of the palmprint image and the iris image are first matched respectively. Then these matching scores are normalized. Finally, the normalized scores are fused to authenticate the identity using the new fusion scheme. The experimental results show that this new scheme can dramatically improve the system performance.