A classification approach to multi-biometric score fusion

The use of biometrics for identity verification of an individual is increasing in many application areas such as border/port entry/exit, access control, civil identification and network security. Multi-biometric systems use more than one biometric of an individual. These systems are known to help in reducing false match and false non-match errors compared to a single biometric device. Several algorithms have been used in literature for combining results of more than one biometric device. In this paper we discuss a novel application of random forest algorithm in combining matching scores of several biometric devices for identity verification of an individual. Application of random forest algorithm is illustrated using matching scores data on three biometric devices: fingerprint, face and hand geometry. To investigate the performance of the random forest algorithm, we conducted experiments on different subsets of the original data set. The results of all the experiments are exceptionally encouraging.