Proprietary Fingerprint Template Evaluation (PFTII)

PFT II has been superseded by PFT III. Visit the PFT III page for more information.

Use this link to see PFT results from 2003 – February 2010

PFTII Overview

 The NIST Proprietary Fingerprint Template Test Phase II is part of an ongoing program to measure the performance of fingerprint matching software by utilizing vendor proprietary fingerprint templates. PFTII NIST will continue to evaluate two of the PFT two index finger datasets, and add a new dataset with ten fingerprints in PFTII. In this new dataset, PFTII will evaluate thumb, index and middle fingers with the following combinations:   plain image vs. plain impressionage, plain image vs. rolled impressionage, and rolled image vs. rolled impressionage.

The sample dataset sizes have been increased to 120,000 subjects. PFTII is still a one-to-one verification evaluation and currently does not have a one-to-many phase. The template extraction time and matching times for three 1K samples with different image widths and heights (368x368, 500x500, 800x800, and 422x1000) will be measured on a uniform hardware and operating system environment. Average template size will also be reported. All new updates can be found in the website FAQ / what's new.

How to Participate

Participants should enroll by completing a PFTII Application form. Participants are limited to submitting an algorithm every 90 days from the completion date of the previous algorithm. Details about the evaluation can be found in the evaluation plan and SDK API, including details on encrypting any submitted software. 

Notes

• The PFTII evaluation is intended to assess the core algorithmic capability of the technology to perform one-to-one verification.

• This evaluation assesses accuracy of end-stage matchers i.e. (the computationally expensive algorithms used in the very last stages of one-to-many Automated Fingerprint Identification System (AFIS) searches). This evaluation is not necessarily representative of a provider's capability to field large-scale identification technologies, because AFIS engineering requires tradeoffs between efficiency, cost, accuracy and other resources, and exploits multi-stage matching techniques to expedite search.

• High Accuracy from the evaluation does not automatically imply a provider has a capability to field a full-scale AFIS installation.

Reports

NISTIR 7821 PFTII report Plain and Rolled Fingerprint Matching with Proprietary Templates  (The report only has results for the first three participants. An updated list of participants and results is available on the PFTII results page .)

NISTIR 7740 Comparison of Confidence Intervals for Large Operational Biometric Data by Parametric and Non-Parametric Methods.

PFTII Links

PFT Homepage
Results
Application
Encryption
Evaluation Plan/API
FAQ
Validation 
ROC Baseline Data