Integrating iris recognition into your product is a significant engineering decision. This guide covers the key specifications, interface options, and performance benchmarks you need to evaluate before selecting an OEM iris recognition module.

1. Understand the Two Types of Iris Modules

Before comparing specs, clarify which type of module fits your use case:

• Capture-only modules — capture and pre-process the iris image, then send it to a host processor or server for matching. Lower cost, simpler hardware, but requires external processing infrastructure.
• All-in-one modules — capture, extract features, and match locally on the module itself. Higher cost, but standalone operation with no server dependency. Suitable for access control, time-attendance, and offline deployments.

2. Key Performance Specifications

False Acceptance Rate (FAR)

FAR is the probability that an unauthorized person is incorrectly accepted. This is your security threshold:

• 1 in 1,000,000 (10⁻⁶) — adequate for physical access control
• 1 in 100,000,000 (10⁻⁸) — financial services, government ID
• 1 in 1,000,000,000 (10⁻⁹) — border control, high-security facilities

HOMSH's Phaselris™ algorithm achieves 10⁻⁹ FAR — the highest security tier.

False Rejection Rate (FRR)

FRR is the probability that a legitimate user is incorrectly rejected. This affects user experience and throughput:

• FRR < 1%: good for general use
• FRR < 0.5%: required for high-traffic entry points
• FRR < 0.1%: HOMSH D-series under standard conditions

Identification Speed

Critical for high-throughput scenarios (factory gates, transit hubs):

• 1:1 verification (user presents ID + scan): typically 0.3–0.5s
• 1:N identification (scan only, match against database): depends on database size and processing hardware
• FPGA-accelerated processing: <0.3s for 1:1, <1s for 1:N up to 10,000 users
• CPU-only processing: 1–3s for 1:1, much slower for 1:N

3. Hardware Interfaces

Match the module's interface to your system architecture:

4. Environmental Ratings

Match the IP rating to your deployment environment:

• IP54 — dust-protected, splash-resistant. Office lobbies, indoor kiosks.
• IP65 — dust-tight, protected against water jets. Outdoor entrances, manufacturing floors.
• IP67 — temporary immersion. Extreme outdoor, mining portals.

5. Illumination: NIR vs Visible Light

All quality iris recognition modules use Near-Infrared (NIR) illumination (700–900nm wavelength). This is non-negotiable for two reasons:

• NIR penetrates the iris more evenly, revealing fine texture patterns hidden to visible light
• NIR is invisible to the human eye — no discomfort, no glare
• NIR performance is consistent across different eye colors (dark brown irises appear similarly textured under NIR)

6. Database Capacity and Matching Architecture

Consider your deployment scale:

• Standalone (<500 users): on-module storage sufficient, no server needed
• Mid-scale (500–10,000): on-module FPGA matching with local database
• Enterprise (>10,000): server-side matching with module acting as capture device

HOMSH modules support both modes. The Qianxin™ FPGA chip enables on-device 1:N matching up to 100,000 templates without server infrastructure.

7. SDK and Integration Support

Evaluate the vendor's software ecosystem before committing:

• Does the SDK support your target OS? (Windows, Linux, Android, RTOS)
• Is the API well-documented? Is sample code available?
• What image formats does the module output? (ISO/IEC 19794-6 compliance is a plus)
• Is there a simulation/evaluation mode for development without hardware?
• What is the support SLA for OEM customers?

8. Liveness Detection

For secure applications, ensure the module includes anti-spoofing (liveness detection). This prevents attacks using printed iris photos or artificial eyes. Look for:

• Pupil dilation response detection
• 3D depth sensing (more robust)
• Multi-spectrum NIR analysis

HOMSH OEM Module Lineup