Performance

ActiveID Biometric Identity Systems outperform allother identity management platforms because of three unique Geometrix capabilities:

  1. FaceVision 3D facial recognition technology, which approaches fingerpring accuracy by using 3D facial shape as a biometric identifier.
  2. ActiveFusion multi-biometric fusion technology, which deeply fuses multiple biometrics such as 3D face, 2D face, and fingerprint modalities to achieve recognition accuracies better than any single biometric.
  3. ActiveID platform functionality and field-proven features such as automated imposter checking, which deliver the only total system solution for large-scale facial identity control.
Example performance results include:

FaceVision 3D Facial Recognition Performance

The most critical and basic performance parameter for any biometric access control system is the curve that defines its false rejection rate. In simple terms, for a given level of assurance that the wrong person will not be let in by mistake, how often is the right person forced to undergo the inconvenience of being falsely rejected?

Geometrix ActiveID systems featuring FaceVision 3D technology outperform any facial alternative in this most critical parameter. The diagram illustrates FaceVision identification algorithm performance as measured by standardized tests conducted for the US Government in early 2005. An "ideal" biometric system would occupy the top left corner of such a diagram.

ActiveFusion Multi-Biometric Fusion Performance

While ActiveFusion performance curves are proprietary, testing to date indicates that:

  1. Fusing 3D face plus 2D face provides a nearly 2X reduction in identification errors compared to the baseline FaceVision 3D product's already industry-leading 3D face performance.

  2. Fusing Fingerprint with 3D face improves equal error rate versus a single fingerprint by up to 40%, while largely overcoming failures that can occur in a fingerprint-only system from causes such as dirty sensors, injured body parts, intentional fingerprint defacement, and spoofing.

System-Level ActiveID Features

The robust and accurate biometric identification performance of ActiveID systems is vital, but just the beginning of the ActiveID platform's real-world problem solving capabilities. ActiveID provides the only total system solution with powerful and complete secondary inspection tools for detailed 3D facial analysis. And here are just two examples of system-level solutions developed with customers for customers:

  1. Prisoner Control:
    Problem: ActiveID systems that control prisoner releases must overcome a high rate of direct challenges to the system, including duplicate IDs and dedicated impostors attempting to exit as a different prisoner.
    Solution: ActiveID Verifier Plus, which performs a 3D facial biometric identity verification for all exiting prisoners, then as an extra precaution, performs an automated one-to-many check and flags close matches for the guard to review.
    Performance: Zero release errors to date after nearly 100,000 prisoner transactions, increased processing rate, and improved staff morale.
  2. High-Security Access Control:
    Problem: Any biometric system will fail to acquire an individual's biometric on occasion, such as a fingerprint sensor failure due to dirty, defaced, or injured fingers.
    Solution: ActiveFusion multi-biometric fusion, which can be set to verify users with extremely high confidence using subsets of the enrolled biometrics, such as "face plus any one of four fingers".
    Performance: Very high security at 0.002% false acceptance rate, without sacrificing convenience.
Want to know more about the world's most advanced identity management platform and technologies? Contact Geometrix today before the next impostor attacks your current identification system.

Copyright © 2007. All rights are reserved.
The true shape of human identity.

Click Graph to Enlarge

In tests of fully automated face recognition, Geometrix ActiveID FaceVision 3D face recognition algorithms achieve superior results, approaching fingerprint levels of performance.

P.J. Phillips, P.J. Flynn, T. Scruggs, K. W. Bowyer, J. Chang, K. Hoffman, J.Marques, J. Min, and W. Worek, "Overview of the Face Recognition Grand Challenge," In Proceedings of IEEE Confernce on Computer Vision and Pattern Recognition, 2005.


Aging / Weight Loss / Beards
The image above is an actual Verification showing a time difference of 18 months. The image on the right is the enrollment template. The image on the left is the probe image where subject lost over 60 lbs and trimmed his beard.


The challenge in facial recognition - varying lighting conditions. The top ranked 2D recognition products perform poorly comparing directional sunlit images to diffusely lit indoor enrollment images. Since 3D face shape is a true invariant, ActiveID systems deliver reliable identity verification under nearly any lighting conditions, including sunlight.