The Science Behind Iris Recognition
From quantum-level phase features to silicon-level ASIC chips — HOMSH owns the full technology stack of iris biometrics.
Core Algorithm
Phaselris™ Algorithm
Phase-Based Feature Extraction
Unlike texture-based approaches, Phaselris™ analyzes the precise phase components of iris fiber patterns using multi-scale Gabor wavelets. This makes the algorithm robust to illumination variations, pupil dilation, and partial occlusion.
1-in-1-Billion Accuracy
Phaselris™ 3.0 achieves a False Acceptance Rate (FAR) of 10⁻⁹ in binocular mode — comparing both eyes simultaneously doubles the security factor. Meets and exceeds NIST IREX and Chinese national standards.
Silicon Implementation
The algorithm is implemented directly in custom ASIC silicon (Qianxin™ chip series), enabling real-time processing at 30ms encoding (QX81) with matching speeds up to 3,200,000 templates/s — 3,200× faster than the Q20 FPGA entry model.
Hardware
Qianxin™ Chip Series
From FPGA prototyping to production ASIC — a complete chip roadmap for every scale of iris recognition system.
| Model | Type | Matching Speed | Encoding | FAR (Single/Dual) | Use Case |
|---|---|---|---|---|---|
| Q20 | FPGA | 10K/s (USB3.0) | 60ms | Single <10⁻⁷ / Dual <10⁻⁹ | Entry-level |
| Q30-200 | FPGA | 40K/s (USB3.0) | 50ms | Single <10⁻⁷ / Dual <10⁻⁹ | Speed-first |
| Q30-75 | FPGA | 4K/s (QSPI) | 50ms | Single <10⁻⁷ / Dual <10⁻⁹ | Area-optimized |
| Q30-50 | FPGA | 12K/s (USB2.0) | 50ms | Single <10⁻⁷ / Dual <10⁻⁹ | 3x Match Engine |
| QX80 | ASIC | 90K/s | 60ms | Single <10⁻⁷ / Dual <10⁻⁹ | Embedded (QFN44/68) |
| QX81 | ASIC | 3,200K/s | 30ms | Single <10⁻⁷ / Dual <10⁻⁹ | Flagship (QFN44/88) |
FPGA
FPGA
FPGA
FPGA
ASIC
ASIC
Technology Roadmap
Recognition Distance Evolution
Standard access control, embedded modules. Products: D36, D50 series. Recognition time ≤1s.
2015–Present
Walk-through gate integration, ATM-distance banking authentication. Prototype tested at 2.8m.
Target: 2025
Border control corridors, stadium access, surveillance integration. Active academic research partnership.
Target: 2027
Why Iris?
Biometric Technology Comparison
| Feature | Iris 👁 | Fingerprint 🖐 | Face 🙂 |
|---|---|---|---|
| Accuracy (FAR) | ✓10⁻⁹ (binocular) | 10⁻⁵ | 10⁻⁴ |
| Anti-Spoofing | ✓Excellent | Moderate | Poor |
| Recognition Distance | Up to 10m (roadmap) | Contact only | ✓3–10m |
| Recognition Speed | ≤1s | ≤1s | ✓≤0.5s |
| Hygiene | Contactless | Touch required | Contactless |
| Stable Over Time | ✓Lifetime stable | Degrades | Ages over time |
| Works with Masks | ✓Yes | No | Partial |
| Dark Environment | Yes (IR) | Yes | Limited |
Integrate Phaselris™ Technology
SDK, chip samples, and technical support available for OEM/ODM integration.