Education · · 3 min

How iris recognition actually works

A near-infrared camera images the iris’ random texture, which is encoded into a compact binary template — not a stored photo — and matched by Hamming distance.

How iris recognition actually works

Iris recognition reads the dense, near-random texture of the iris — the colored ring around the pupil. Unlike a face or a fingerprint, that texture is fully formed in early childhood — by roughly ten months of age — and then stays remarkably stable over a lifetime, which is what gives the iris its high discriminative power.

A dedicated camera images the eye under near-infrared light. The system localizes the iris, masks out eyelids, lashes and specular reflections, and “unwraps” the ring into a normalized strip so that pupil dilation does not change the result.

That strip is filtered and encoded into a compact binary template — often called an IrisCode — rather than stored as a photograph. Two templates are compared by counting the fraction of bits that differ (a Hamming distance); a low distance means a match.

Because the comparison is a fast bitwise operation, one iris can be checked against very large galleries in milliseconds. NIST’s independent IREX program tracks how well commercial matchers do this across standardized, interoperable iris imagery.

Source: NIST — Iris Exchange (IREX)

All updates