MinuCode: A Fixed-Value Representation of Fingerprint Minutiae for Biometric Cryptosystem

摘要

This paper proposes a new fingerprint verification algorithm, MinuCode, which represents minutiae information in fixed values that can tolerate distortions during different fingerprint image scanning processes. The fixed-value representations of minutiae make MinuCode a most suitable candidate to serve as a basis for fingerprint biometric cryptographic systems. To deal with the fingerprint translation, rotation and non-linear distortion, MinuCode first utilizes a minutiae-centered encoding mechanism which eliminates the need for reference core point determination, and quantizes the location relationship between each pair of minutiae to a fixed value. Furthermore, missing or additional spurious minutiae are handled with a threshold-based* matching algorithm. Experimental results on the public domain database showed the MinuCode scheme can achieve recognition accuracy similar to the basic method of traditional fingerprint recognition. Thus, together with the fixed-value representations of minutiae, MinuCode is a most suitable and practical fingerprint matching scheme for supporting fingerprint-based biometric cryptosystem.