BaselineWatch
Desk
Your fleet and your work Fleetyour releases and the hardware they run onTriagewhat needs a decisionWatchlistthe products and platforms you runSaved searchesthe lists you come back to
This instance Sourceswhat is read, how often, under what termsDigestthe daily or weekly summary, and alerts
How it is drawn Appearancethe look and the colour schemeBrandingthe mark this instance wears
Owner sign-in

10.4 Conformance to accuracy specifications

NIST PIV and digital identity SP 800-76-2 against SP 800-76-1
All changes in this version
added 10 Minimum accuracy specifications part of 10

Title changed

Before · SP 800-76-1

After · SP 800-76-2

Conformance to accuracy specifications

Control changed

Before · SP 800-76-1

After · SP 800-76-2

The false match rate (FMR) requirements shall be assured by submitting biometric comparison algorithms to the test and calibration programs given in Table 17 . Those programs shall provide the algorithm developer with a tabulation of false match rate vs. threshold calibration.

Table 17 – Example performance test and threshold calibration programs Modality Authentication Test + Calibration Program Status Fingerprint minutia matching Off-card MINEX III This program was formerly known as Ongoing MINEX. It includes an interoperability component (Level 1 certification) and an operational support component (Level 2 certification). On-card MINEX IV This program follows the [MINEXII] protocol to implement the Level 1 and 2 components described in MINEX III. Iris image matching Off-card IREX VIII This program follows the [IREX-I] test ensuring correct generation and matching of PIV Card (Image Type 7) standard images. Face image matching Off-card Agencies may reference recent test results from any source, or take a vendor-recommended value. Examples of such tests are [FACEPERF].

The test measurements are typically obtained by running algorithms on commodity PC hardware. Thereafter, the algorithm provider or integrator shall provide documented attestation that: ― All components of the recognition software (including template generation and comparison algorithms) are functionally identical to those submitted to the recognized test and calibration program. The use of recognition algorithms on other platforms, such as wall mounted embedded processors, is allowed. The algorithm provider shall submit the same software to the test program wherever it is ultimately installed. ― All instances of the fielded comparison algorithms are configured with an operating decision threshold that is at least as strong as that established in FMR vs. threshold calibration. Agencies might require inspection of source code and institute appropriate controls to ensure that the source code is indeed that installed in deployed equipment. Additionally, Agencies could elect to conduct a biometric performance test to confirm the hypothesis that the FMR is conformant to the specification of Section 10.3 .