Biometric face image quality standard draft released along with assessment tool

The Open Source Face Image Quality (OFIQ) assessment software library and the Draft International Standard it is a reference implementation for are now available on GitHub.

OFIQ can be used to identify images that will be difficult for face biometrics systems to process due to low image quality.

ISO/IEC 29794-5 is the standard for face image data within the biometric sample quality family of standards. It is applicable to enrollment for biometric passports and the biometric kiosks used for the EU’s EES. The sixth working draft went out to working group members for comments last February.

“The standard will define a unified quality score for a face image and numerous quality components, which provide accurate information such as the correctness of the pose or the sharpness of the face image (and many others),” says a LinkedIn post from ISO/IEC JTC 1/SC 37 Biometrics.

The Draft International Standard is posted to the GitHub repository as document ISO/IEC JTC 1/Sc 37/WG 3 N 1511, along with the open source code for OFIQ.

For its part, NIST has adapted its Face Analysis Technology Evaluation (FATE) Part 11: Face Image Quality Vector Assessment to better align with ISO 29794-5:2024.

Image quality was also judged to be the main challenge to accurate biometric matching in DHS’ 2022 Biometric Technology Rally. A representative of DHS’ biometric and identity technology center told a federal Identity and Access Forum meeting that facial recognition is successfully matching 95 percent of people correctly, but 75 percent of observed “errors are due to camera image capture problems,” according to a Secure Technology Alliance announcement.

Article Topics

biometric data quality  |  biometrics  |  face biometrics  |  facial recognition  |  image analysis  |  ISO standards  |  Open Source Face Image Quality (OFIQ)  |  standards