An AFIS Using Fingerprint Classification

Fingerprints have been used as biometrics for personal identification or verification since a century ago. Although no exactly the same fingerprint from distinct identities was found, a perfect system for automatic fingerprint identification does not exist. This paper implements an automatic fingerprint identification system (AFIS) with the use of fingerprint classification and minutiae pattern matching. A minutiae pattern is composed of various minutiae extracted from a fingerprint image. Each minutia is represented by its relative location, a type: ending or bifurcation, and the ridge direction. Our system is tested on a database composed of 308 300 × 300 left index fingerprint images contributed by 77 persons via a Veridicom FPS110 reader. A 80% recognition rate is achieved. Each matching takes 5 ∼ 9 seconds CPU time on a PC with K6-400 CPU and 128 MB SDRAM running a Windows 2000 OS.