IMAGE QUALITY ASSESSMENT FOR FAKE BIOMETRIC DETECTION: APPLICATION TO FINGER-VEIN IMAGES.

A biometric authentication system is most widely accepted and used across all fields in real life. Among all the biometric techniques, most of the extrinsic biometric traits (e.g., fingerprint and face) are susceptible to spoof attacks. Finger-vein biometric unlike the others provide a secure mechanism for authentication due to its hidden nature. However the finger-vein biometric system is to some extent vulnerable to spoofing and other indirect attacks. The work presents a software based liveness detection method to address the vulnerability issue. The proposed work focuses on Image Quality Assessment approach where a set of 10 Quality measures are extracted from a single image to recognize the forged samples from the real ones. It is more efficient, less intrusive and fast compared to the other state-of-the art works which makes it feasible for real-time applications. The work uses the publically available database to evaluate the performance of the system which provides better FPR and FNR rates.