ir_7112 - Studies of Plain-to-Rolled Fingerprint Matching...

Info iconThis preview shows pages 1–3. Sign up to view the full content.

View Full Document Right Arrow Icon
Studies of Plain-to-Rolled Fingerprint Matching Using the NIST Algorithmic Test Bed (ATB) NISTIR 7112 Stephen S. Wood Charles L. Wilson April 2004
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Studies of Plain-to-Rolled Fingerprint Matching Page 1 of 47 Studies of Plain-to-Rolled Fingerprint Matching Using the NIST Algorithmic Test Bed (ATB) Stephen S. Wood & Charles L. Wilson (from the National Institute of Standards & Technology) EXECUTIVE SUMMARY A series of fingerprint matching studies have been conducted on an experimental laboratory system called the Algorithmic Test Bed (ATB). The ATB is a system used to test the functional characteristics of the Automated Fingerprint Identification System (AFIS) component of the Federal Bureau of Investigation’s (FBI) Integrated Automated Fingerprint Identification System (IAFIS). It was supplied by Lockheed Martin a , the vendor of IAFIS, whose technology it shares. The ATB includes a built-in gallery of nearly 1.2M subjects, effectively a 3% representative sample of the FBI’s criminal master file, with some room for additions. The ATB provides broad control over its operating modes and set points. The ATB, like IAFIS, was originally designed to identify (and verify) subjects using submitted (predominantly rolled) fingerprint cards. The database against which the submitted cards were to be matched was likewise predominantly composed of rolled images from fingerprint cards. As demand rose for the use of plain (or flat or slap) images, ascertaining the operational characteristics of this matcher with plain submissions became important. These studies were organized in three major groups. Studies of plain-to-rolled matching as a function of data source made up the first group. Studies of plain-to-rolled matching using large sets of operational data made up the second. Studies of plain-to-plain matching made up the third. The studies answered nine major questions: (1) how accurate is plain-to-rolled identification? is it less accurate than rolled-to-rolled? Failure to identify (1-TAR) can be 2% or less. Accuracy can be as good as with rolled prints of average quality. (2) how accurate is plain-to-rolled identification with fewer than ten fingers? Accuracy with fewer than ten fingers can exceed accuracy with ten fingers, but performance penalties can be expected (see next question). (3) are performance penalties incurred by adopting plain-to-rolled identification? if penalties are incurred, how great are they? Performance (throughput) deteriorates markedly as the amount of data is reduced. Using plain rather than rolled fingerprints in 10-finger or 8-finger matching can be expected to result in a three-fold reduction in throughput. Reducing the number of fingers from ten to two can be expected to reduce throughput by a factor of ten or more. These effects are independent of one another, and thus multiplicative.
Background image of page 2
Image of page 3
This is the end of the preview. Sign up to access the rest of the document.

This note was uploaded on 09/27/2010 for the course 6511 5487 taught by Professor Chaohue during the Spring '10 term at Mackenzie.

Page1 / 48

ir_7112 - Studies of Plain-to-Rolled Fingerprint Matching...

This preview shows document pages 1 - 3. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online