What is Biometric Portrait Consistency?

What is Biometric Portrait Consistency?

Biometric Portrait Consistency evaluates whether two or more portrait images printed on the same document appear to represent the same individual. On supported biometric passports and identity documents, the check compares the primary portrait against a secondary or ghost portrait, or against another portrait representation contained on the same document.

The output is a diagnostic result — typically CONSISTENT, INCONSISTENT, NOT_ASSESSABLE, or NOT_APPLICABLE — reported inside the Veridexa fraud detection report as additional technical evidence.

Why it matters

Portrait substitution is one of the oldest and most common forms of identity document fraud. An attacker who successfully replaces the primary portrait of a genuine document with their own photograph often forgets, or is unable, to update the ghost portrait or other secondary representations printed elsewhere. Comparing the multiple portraits on the same document therefore surfaces a specific class of tampering that OCR and text-only checks cannot see.

Biometric Portrait Consistency is not a replacement for image forensics, MRZ validation, or security-feature analysis. It is one more independent line of evidence that strengthens the overall fraud detection report on supported biometric documents.

Supported documents

Biometric Portrait Consistency is currently applicable to supported documents that carry two or more visible portrait representations, including:

• Biometric passports following ICAO Doc 9303 (TD3) layouts
• National identity cards in ID-1 format
• Residence permits in ID-1 format
• Driving licences with a ghost or secondary portrait

Documents that contain only a single portrait, or documents where a portrait cannot be reliably located, are treated as NOT_APPLICABLE or NOT_ASSESSABLE — no fraud finding is generated in those cases.

How Veridexa analyzes portrait consistency

For supported document types, Veridexa automatically locates the expected portrait regions using known document layouts and extracts the primary and secondary portrait crops from the submitted image or PDF page. Non-face regions and low-quality crops are rejected so that only reliably usable portraits are compared.

Retained portraits are normalized to a consistent orientation and scale, converted into structural representations, and compared against each other to produce a similarity indicator. The similarity indicator is then mapped into one of four discrete outcomes:

• CONSISTENT — the portraits are technically consistent with representing the same person.
• INCONSISTENT — the portraits differ enough that they may not represent the same person.
• NOT_ASSESSABLE — one or more portraits could not be reliably evaluated (for example, low resolution, glare, or partial occlusion).
• NOT_APPLICABLE — the document type does not include multiple portraits, or none could be located.

The result appears in the Veridexa report under a dedicated Biometric Portrait Consistency section as neutral, diagnostic evidence.

How it strengthens document fraud detection

A single portrait tells you what someone looks like; two or more portraits printed on the same document tell you whether the document agrees with itself. That is a powerful additional signal in an evidence-based fraud detection workflow, complementing:

• OCR Intelligence and MRZ validation on printed fields
• Image Forensics on the pixel content of the page
• Security Features and Document Structure analysis
• QR Validation and Digital Signature Validation, where present

Reviewers can use Biometric Portrait Consistency as one more explainable signal alongside the rest of the Veridexa report, particularly when triaging suspected portrait substitution.

Limits and what it does not claim

• Biometric Portrait Consistency compares portraits contained within a single submitted document. It is not an identity-matching or face-recognition service against external databases.
• It does not access the electronic chip, NFC, or any biometric data outside the submitted document image or PDF page.
• On Veridexa, the result is published as diagnostic evidence and does not, on its own, change the final fraud decision.
• Veridexa does not claim biometric-grade or perfect accuracy for this check, and does not describe it as superior to any other provider or approach.

Veridexa Analysis

On Veridexa, Biometric Portrait Consistency is one of several analytical capabilities powering document fraud detection. It is available for supported biometric passports, national ID cards, residence permits, and driving licences, and appears alongside OCR Intelligence, Image Forensics, Security Features Analysis, Document Structure Analysis, QR Validation, and Digital Signature Validation in every applicable report.

To see the full analysis suite for a specific document, run a Veridexa Fraud Detection analysis and open the report — the Biometric Portrait Consistency section will appear whenever the document contains multiple portrait representations.

Frequently Asked Questions