Enhanced Biometric Capture Validation with SDK Match Score & Face Quality Checks

[Pragya279]

User Story: Enhanced Biometric Capture Validation with SDK Match Score & Face Quality Checks

As a Registration Operator
I want the system to show the biometric match score from the SDK along with the capture quality score, and automatically detect if the captured face image is non-smiling or low-quality
So that I can ensure the biometric is both high-quality and suitable for matching, ensuring accurate identification and a better applicant experience.

---

Problem / Need

Capturing a technically good biometric is not sufficient—identification accuracy depends on matchability (how well the biometric will match future samples) and image quality (clarity, lighting, pose, expression).

• Operators currently see only quality scores.

• They do not see matchability indicators, which help assess whether the captured biometric will perform well in deduplication or authentication.

• For face capture, smiling, inconsistent lighting, tilt, or blur can make photos unusable.

• The system needs to automatically detect such issues and guide operators for better captures.

This feature closes these gaps by combining Quality Score, Match Score, and Face Expression/Quality Validation.

---

Feature Description

1. SDK Match Score Display

• After capturing biometrics (face, fingerprint, iris), the system sends the captured sample to the configured SDK (e.g., face SDK, fingerprint matcher).

• SDK returns a Matchability Score such as:

  • Match Score (0–1 or 0–100)

  • Matching Confidence Level (Low/Medium/High)

  • Liveness or anti-spoof checks (if supported)

• The system displays this along with the standard quality score.

Output Seen by Operator:

• Quality Score (clarity, sharpness, positioning)

• Match Score (SDK-based matchability)

• Visual indicator:

  • Green: Good matchability

  • Yellow: Acceptable

  • Red: Poor / unusable

---

2. Non-Smiling Face Detection

• Using face analytics from the SDK, the system detects:

  • Smile or unacceptable expressions

  • Mouth open

  • Eye closure / blink

  • Head tilt or rotation

• If smiling or expressive face is detected:

  • UI shows:
    “Neutral expression required. Please recapture the photo.”

---

3. Enhanced Face Image Quality Checks

For face captures, the system automatically validates:

• Lighting uniformity

• Background consistency

• Sharpness & focus

• Pose correctness (frontal, straight eyes, no tilt)

• Occlusions (hair, mask, glasses glare)

If the face capture is poor, the system prompts the operator to retake it.

---

4. Display of Combined Validation Results

The capture screen will show:

Parameter | Displayed to Operator -- | -- Biometric Quality Score | Numeric + Pass/Fail Match Score | Numeric + Matchability rating Expression Check | Pass/Fail (e.g., “Neutral Face Required”) Photo Quality Indicators | Highlights issues: Blur / Lighting / Pose

Operators see all validations in one place.

---

5. Configurable Thresholds

• Quality thresholds

• Minimum match score

• Expression rules (strict/relaxed)

• Recapture enforcement (mandatory or override-allowed)

---

6. Storage in Packet

• Store the quality score, match score, and face analysis metadata inside biometric packet metadata for audits and analytics.

---

Acceptance Criteria

AC1: System calculates and displays both quality score and SDK match score after capture.

AC2: System detects smiling or expressive faces and blocks capture until a neutral expression is provided.

AC3: System validates photo quality (lighting, focus, pose, occlusion) and prompts operator for correction.

AC4: System blocks operator from proceeding if match score < threshold (unless override enabled).

AC5: All scores and metadata are stored in the biometric packet.

AC6: Thresholds for match score and expression rules are configurable.

AC7: UI shows clear messages for each error or poor condition.

---

Example User Flow

1. Operator captures applicant’s face photo.

2. System calculates:

   • Quality score = 72 (Pass)

   • Match score = 48 (Below threshold 60 → Fail)

   • Expression: Smiling → Fail

   • Pose: Slight head tilt → Warning

3. System displays:

   • ❌ Matchability too low. Please recapture.

   • ❌ Neutral expression required.

   • ⚠️ Head slightly tilted. Adjust camera.

4. Operator recaptures face.

5. New results:

   • Quality = 85 (Pass)

   • Match score = 78 (Pass)

   • Neutral expression (Pass)

   • Pose correct (Pass)

6. Operator proceeds.