Fingerprint classification and matching using a filterbank

Accurate automatic personal identification is critical in a variety of applications in our electronically interconnected society. Biometrics, which refers to identification based on physical or behavioral characteristics, is being increasingly adopted to provide positive identification with a high degree of confidence. Among all the biometric techniques, fingerprint-based authentication systems have received the most attention because of the long history of fingerprints and their extensive use in forensics. However, the numerous fingerprint systems currently available still do not meet the stringent performance requirements of several important civilian applications. To assess the performance limitations of popular minutiae-based fingerprint verification system, we theoretically estimate the probability of a false correspondence between two fingerprints from different fingers based on the minutiae representation of fingerprints. Due to the limited amount of information present in the minutiae-based representation, it is desirable to explore alternative representations of fingerprints. We present a novel filterbank-based representation of fingerprints. We have used this compact representation for fingerprint classification as well as fingerprint verification. Experimental results show that this algorithm competes well with the state-of-the-art minutiae-based matchers. We have developed a decision level information fusion framework which improves the fingerprint verification accuracy when multiple matchers, multiple fingers of the user, or multiple impressions of the same finger are combined. A feature verification and purification scheme is proposed to improve the performance of the minutiae-based matcher.