Skip to Main Content
Biometric identification has lately attracted attention because of its high convenience; it does not require a user to enter a user ID. The identification accuracy, however, degrades as the number of the enrollees increases. Although many multimodal biometric techniques have been proposed to improve the identification accuracy, it requires the user to input multiple biometric samples and makes the application less convenient. In this paper, we propose a new multimodal biometric technique that significantly reduces the number of inputs by adopting a multihypothesis sequential test that minimizes the average number of observations. The results of the experimental evaluation using the NIST BSSR1 (Biometric Score Set - Release 1) database showed its effectiveness.
Date of Conference: 6-9 Dec. 2009