ABIS & Deduplication

Automated Biometric Identification Systems perform large-scale 1:N searches to resolve identities and detect duplicates.

Overview

ABIS platforms store and search biometric templates at national or global scale. Deduplication ensures each person has only one identity record.

How it works

  1. Capture biometric samples during enrollment.
  2. Extract features and store templates.
  3. Run 1:N searches to detect matches or duplicates.
  4. Adjudicate hits and maintain watchlists.

Common use cases

  • National ID enrollment
  • Border and visa vetting
  • Civil or criminal watchlists

Strengths and limitations

Strengths: Scales to millions; prevents multiple identities.
Limitations: Infrastructure cost; privacy and governance.

Key terms

  • 1:N search: Matching a probe against all records.
  • Deduplication: Removing duplicate identities in a database.

References

Vendors using ABIS & Deduplication

Latest Data Cards

  • Data Card

    Innovatrics ranks first on four NIST latent fingerprint accuracy measures

    2026-07-28CC-BY-4.0fingerprint-recognitionabis-dedupinnovatrics

    The latest NIST Evaluation of Latent Friction Ridge Technology placed the Innovatrics submission first across four accuracy measures: two hit-rate measures and two false negative identification rate measures.

    • The innovatrics+0014 submission placed first for two hit-rate measures and two false negative identification rate measures.
    • It recorded a 99.3 percent rank-1 hit rate without similarity-score filtering, and 98.9 percent at rank-100 at a 10 percent false positive identification rate.
    • On a Department of Defense set of over 5,257 latent probes against about 1.6 million reference profiles, it returned a 4.85 percent false negative identification rate at a 0.01 false positive rate.
  • Data Card

    ROC posts fastest latent fingerprint search in NIST evaluation

    2026-07-17CC-BY-4.0fingerprint-recognitionabis-dedup

    ROC reported the fastest search speed in NIST's Evaluation of Latent Fingerprint Technologies, alongside a first-place rank-1 result on one FBI dataset subset.

    • ROC states its algorithm searched roughly 64 times faster than the mean of the top five vendors.
    • It reports a first-place rank-1 result on the Extended Feature Set subset of the FBI-provided solved dataset.
    • Testing used FBI and Department of Defense investigative datasets across a field of 22 vendors.
  • Data Card

    FBI issues third market-research RFI for biometric matching algorithms

    2026-06-22CC-BY-4.0abis-dedupfingerprint-recognitionfacial-recognitioniris-recognition

    The FBI issued a third market-research request for biometric matching algorithms as it surveys options for the Next Generation Identification environment.

    • The RFI asks vendors for information on biometric algorithms, response times, accuracy, scalability, transition approach, maintenance, licensing, and secure build practices.
    • The FBI's Criminal Justice Information Services division operates the Next Generation Identification system used for large-scale law-enforcement biometric matching.
    • Responses are due July 15, 2026, and the notice is market research rather than a procurement commitment.

Frequently Asked Questions

What does an ABIS do?
It searches biometric databases to identify individuals and flag duplicate enrollments across large populations.
What’s the difference between identification and verification?
Identification is 1:N (who is this?); verification is 1:1 (is this person who they claim?). ABIS primarily supports 1:N at scale.
How are false positives managed?
Systems tune thresholds and use adjudication workflows; quality checks and multi‑modal fusion reduce spurious hits.