Iris-biometric comparators: Exploiting comparison scores towards an optimal alignment under Gaussian assumption

Iris-based systems guarantee a level of security and identity protection that is unparalleled by any other biometric. In past years numerous iris recognition algorithms have been proposed revealing impressive recognition rates. While the vast majority of research is focused on extracting highly discriminative feature vectors potential improvements in biometric comparators are commonly neglected. In this paper a new strategy for comparing binary biometric templates, in particular iris-codes, is presented. Instead of optimally aligning two iris-codes by maximizing the comparison score for several bit shifts utilizes the total series of comparison scores, avoiding any information loss. The soundness of the approach, which requires marginal additional computational effort, is confirmed by experiments applying two different iris-biometric feature extraction algorithms.