Fingerprint matching based on global minutia cylinder code

Although minutia set based fingerprint matching algorithms have achieved good matching accuracy, developing a fingerprint recognition system that satisfies accuracy, efficiency and privacy requirements simultaneously remains a challenging problem. Fixed-length binary vector like IrisCode is considered to be an ideal representation to meet these requirements. However, existing fixed-length vector representations of fingerprints suffered from either low distinctiveness or misalignment problem. In this paper, we propose a discriminative fixed-length binary representation of fingerprints based on an extension of Minutia Cylinder Code. A machine learning based algorithm is proposed to mine reliable reference points to overcome the misalignment problem. Experimental results on public domain plain and rolled fingerprint databases demonstrate the effectiveness of the proposed approach.