BioKey Matching Algorithm

1.Biokey is a high speed and accurate 1:M Fingerprint Matching algorithm catered largely for software developers. When using Biokey for 1 - M identification ( 2000-6000 fingerprints), there is no requirement to enter a name or a PIN number. Identification can be carried with ease within 1-5 seconds ( Minimum hardware requirement of Pentium III, 900MHz PC). Biokey has the following special features

2.Biokey SDK allows for easy integration to end user applications. Through various image files, it can support many fingerprint sensor types ( with resolution of >= 300DPI).

3.Biokey algorithm is able to filter off noises, ridge ruptures and stuck ridges from fingerprints. Even for fingerprints that are of bad quality ( dirty, cuts, scars, dry, wet or damaged), the algo is able to extract accurate minutae points.

4.Biokey Matching algo allows for 0 to 360 degree rotation of live capture for identification or verification. With state-of-the-art techniques, the match can still be conducted at high speed despite the different orientation angles of the finger ( Approx Matching Speed = 3000 Match/sec) even for fingers with low minutae count(<=10), a fingerprint normally would have >= 15 minutae counts.

5.Biokey Matching algorithm does not require complete minutae points such as (core, delta, etc), it can successfully identify based on several essential or critical minutae points.

6.Biokey categorization method ( Fingerprints are divided mainly into 5 categories: ( Arch, left loop, right loop, tented arch, whorl), if categorization is used in the first instance, the matching speed will be greatly increased.

7.Biokey Matching algorithm source code is simple, only 200Kbytes of program space is required, therefore it is easily accommodated into any PC hardware and software programs.

..From 4 types of Sensor ( YLC, DFR200, U.ARE.U, Authentec ) we obtained 2000 fingerprints to test the performance of the Biokey Matching algorithm ( For Each Sensor, we obtained 500 fingerprints), each fingerprint obtained is than cross matched with the rest of the 2000 fingerprints, about 4,000,000 matches were executed, the following are the performance results:

Fingerprint template size 256 or 1024 bytes
Rotation 0 - 360 degrees
FAR <= 0.001%
FRR <= 2.0%
Enrollment time 0.5 Seconds
Matching Speed 2000-4500 Matches/sec
Livescan Image > = 300DPI