What Is Iris Recognition? How Does It Work?

Look into a mirror. Lean in close.

That colored ring around your pupil does more than give your eyes their shade. It holds a tangle of ridges and pits and threads that formed before you were born. Nobody else on Earth carries the same one.

Not your parents. Not your identical twin. Not even your own other eye.

That’s the finding of Cambridge scientist John Daugman, whose math still powers most iris systems today.

This tiny pattern has made iris recognition one of the most trusted ways to prove who you are. It screens travelers at every UAE border point and unlocks Apple Vision Pro. It also backs the identity of more than 1.38 billion Aadhaar holders in India.

So what is iris recognition? And how does a camera turn your eye into a password? Let’s walk through it.

What is iris recognition?

Iris recognition is a way to identify a person from the patterns in their iris. It’s also called iris biometrics or eye recognition, and it’s one of many types of biometrics in use today.

Here’s the short version. A special camera photographs your eye. Software turns the pattern into a digital code and compares it with codes stored earlier.

The iris itself is the colored ring of muscle that opens and closes your pupil. It sits behind the clear cornea. That keeps it protected but still easy to see, a bit like a watch face under glass.

Why your iris is one of a kind

Color isn’t what matters here. Texture is.

Zoom in on any iris and you’ll find a whole landscape. There are small pits called crypts and rings of folds called furrows. There’s a zigzag border called the collarette and a scatter of freckles.

According to Daugman’s research, this texture starts forming in the third month of pregnancy. It’s mostly finished by the eighth.

Here’s the fascinating part. Random events shape those details as tissue folds and tears in the womb. Your genes have little to do with it.

That’s why identical twins have different iris patterns. To a computer your left and right eyes look as different as two strangers.

Two jobs: verify or identify

Iris systems do one of two jobs, much like other biometric authentication methods.

Verification asks a simple question. Are you who you claim to be? This is also called iris authentication.

Unlocking a headset with your eyes is a good example. The device only checks you against your own stored template.

Identification is a far bigger task. The system searches your iris against thousands or even millions of records to work out who you are. Border control and national ID programs do this every day.

How does iris recognition work?

An iris scan feels instant. You glance at a camera and a second later the door opens.

Behind that second sit five quick steps. Most systems still follow the method John Daugman published in the 1990s. The FBI’s Biometric Center of Excellence describes the same steps in its own guide.

Step 1: Capture the eye

The camera lights your eye with near infrared light between 700 and 900 nanometers. You can’t see it.

This light works from up to about a meter away. It also reveals rich texture in very dark brown eyes, where a normal photo sees only a flat dark disc.

Step 2: Find the iris

Next the software finds two circles. One is the edge of the pupil and the other is the outer edge of the iris. Then it masks out eyelids and lashes and any glare.

This step matters more than it sounds. The FBI guide warns that sloppy work here lets noise creep in. And noise is what makes matches fail.

Step 3: Unwrap the ring

Your pupil grows and shrinks with light. That stretches the iris like a rubber band.

So the software “unrolls” the ring into a flat rectangle. The pattern then stays comparable whether your pupil is wide open in a dim room or tiny in bright sunlight.

Step 4: Turn the pattern into a code

Mathematical filters scan the flat texture and record how it changes from point to point. The result is an IrisCode of 2,048 bits. That’s just 256 bytes plus a set of mask bits that mark the hidden parts.

The whole code is smaller than this paragraph. And it’s a string of ones and zeros rather than a photo of your eye.

Step 5: Compare the codes

Finally the system lines up your new code against a stored one. It counts how many bits disagree. Experts call this share the Hamming distance.

Two different eyes disagree on about half their bits, just like two runs of random coin flips. A code from the same eye disagrees on far fewer. Daugman suggests treating anything under about 32% disagreement as a match.

The math is simple, so it’s blazing fast. Even a 300 MHz processor from decades ago could run about 100,000 comparisons every second.

Iris scan vs retina scan: not the same thing

Movies love to blur the two. In reality they read completely different parts of your eye.

A retina scan maps the blood vessels at the very back of your eye. You have to press close to an eyepiece while light shines deep inside, as SecureIDNews explains. An iris scan simply photographs the front of your eye from a comfortable distance.

	Iris scan	Retina scan
What it reads	Texture of the colored ring	Blood vessels at the back of the eye
How close you get	Up to about a meter away	Eye held close to an eyepiece
How it feels	Like having a photo taken	More intrusive
Health effects on the pattern	Texture stays stable	Diabetes or glaucoma can change it
Real world use	Borders, ID programs and devices	Rare outside high security sites

That gap in comfort is a big reason the iris won the commercial race. For a deeper look see our breakdown of the differences between iris recognition and retina scanning.

How accurate is iris recognition?

Very. But “very” needs numbers.

The most respected independent test is IREX. It’s run by the US National Institute of Standards and Technology (NIST). In the current round, called IREX 10, NIST tests vendor algorithms on large sets of real images the companies never see.

One top performer is Neurotechnology’s algorithm. NIST asked it to find people in a database of 500,000 enrolled people. With false alarms held to 1 in 100 searches, here’s how often it missed the right person:

Setup	Missed matches per 1,000 searches	Missed at a stricter 1 in 1,000 false alarm setting
Both eyes	2.0	2.3
One eye	5.3	6.4

Two things jump out. Using both eyes cuts misses by more than half. A tenfold stricter false alarm setting barely moves the miss rate, which points to a very stable system.

NIST also tested twins using 5,078 iris images from 691 twins. It’s a direct check on the claim that even identical twins carry different irises.

Built for huge databases

Accuracy at small scale is easy. The real test of iris identification comes when a system must compare every new person against everyone already enrolled.

The UAE’s border system ran 62 trillion iris comparisons over ten years. By early 2013 it had stopped 351,318 previously deported people from coming back in.

India’s Aadhaar went further still. At the height of enrollment in 2014 it ran about 600 trillion comparisons a day, according to Daugman’s history of the field.

Where you’ll meet iris scanning today

You may have used iris recognition without a second thought. Here are the places it shows up most.

Airports and borders

The iris went to work at borders early. Soon after 2000 Amsterdam’s Schiphol Airport and ten UK terminals let registered travelers skip the queue with a glance. The UAE followed in 2003.

Today in the US the trusted traveler service CLEAR lets members pass security with a fingerprint or an eye scan. Many airports have since added face scanning too, as these airport facial recognition examples show.

National ID programs

India’s Aadhaar is the giant here. It collects ten fingerprints and both irises from each person. The iris quietly solves a real problem.

Manual laborers often have worn fingerprints. Some people have no fingers at all.

Since late 2023 anyone who can’t give fingerprints can enroll with an iris scan alone. Nearly 2.9 million people with missing biometrics had Aadhaar numbers by then.

Phones and headsets

Samsung put an iris scanner on its Galaxy flagships in the late 2010s. It later dropped the feature for fingerprint and face unlock. The iris came back in 2024 with Apple’s Optic ID on Vision Pro.

Optic ID reads your iris with near infrared light and the headset’s eye cameras. Apple says the chance a random person could unlock your device is less than 1 in 1,000,000. Your iris data stays encrypted on the device and is never backed up to iCloud.

Proof of personhood

The newest and most debated use comes from World. It’s the project co-founded by OpenAI’s Sam Altman. Its silver Orb scans your iris to prove you’re a unique human on an internet filling up with AI bots.

By July 2026 more than 18 million people had verified with an Orb. The same report notes suspensions and investigations in several countries. That brings us to the harder questions.

The limits and the debates

No biometric is perfect. Some of the iris’s weak spots are technical while others are about trust. Experts still disagree on how much they matter.

For a full rundown see our guide to the advantages and disadvantages of iris recognition.

Can an iris scanner be fooled?

Sometimes yes. In 2017 Germany’s Chaos Computer Club unlocked a Samsung Galaxy S8 with a simple trick.

They photographed the owner’s eye in night mode and printed it life size. Then they laid a contact lens on top to fake the curve of a real eye.

The industry’s answer is liveness detection. It checks for signs of a living eye such as pupil reaction and depth. Researchers even run open contests like LivDet-Iris to see which methods catch fakes best.

Supporters say modern systems have closed the old gaps. Skeptics point out that attackers keep evolving too.

Does your iris really never change?

The classic claim is “enroll once for life.” A 2012 Notre Dame study challenged it. Match failures rose over three years of repeat scans.

Then a 2014 MITRE study looked at about 3.5 million US military iris images. It found no clear link between time and matching errors. Only a few people declined, likely due to medical conditions.

So the question is still open. Medical events may matter more.

A 2019 study in Scientific Reports followed 132 cataract patients. Match scores dropped from about 93% before surgery to about 75% after it. Some people may need to enroll again.

Who holds your iris and why?

This is where opinions split sharply. Unlike a password you can’t reset your iris if a database leaks.

Some systems keep your data at home. Apple says Optic ID data never leaves your device. National programs and projects like World handle iris data at a much larger scale.

That scale draws questions from regulators and privacy groups. It has also led some authorities to suspend or investigate World.

Supporters reply that iris IDs give millions of people their first official identity. They also block fraud and spare travelers long queues.

Both sides have real evidence behind them. Where you land depends on how you weigh convenience and inclusion against control over your own body data.

Is the light safe?

Scanners use near infrared light that you can’t see. The FBI guide notes this light can’t cause photochemical damage. It calls the risk from a single LED remote, though badly designed devices with many LEDs could cause harm.

That’s one reason reputable systems follow eye safety standards.

What’s next for iris recognition?

Iris recognition technology is more than 30 years old. Yet it feels like it’s entering a second life. Three trends stand out.

More competition, better algorithms. IREX 10 had drawn 147 submissions from 35 companies by May 2026. That count comes from Neurotechnology, one of the participants. More rivals usually mean faster gains.

Your eyes as the login for headsets. Mixed reality headsets already track your eyes to work. So the cameras are there anyway. Optic ID shows how that hardware can double as a password, and smart glasses may follow.

Proving you’re human. AI bots keep getting harder to spot online. Iris scans are now pitched as proof that a real person is behind an account. World is already piloting this with Tinder in Japan.

How regulators respond will shape how far that idea spreads.

Frequently asked questions

What is an iris scan?

An iris scan is a photo of your eye taken in near infrared light. Software turns your iris texture into a short code. That code can be matched later to confirm who you are.

What is iris recognition used for?

It checks travelers at borders and airports. It confirms identity in national ID programs like Aadhaar. It also unlocks devices such as Apple Vision Pro and now helps prove you’re human online.

Does iris recognition work with glasses or contact lenses?

Usually yes. Clear glasses and normal contacts rarely cause trouble. Strong reflections or patterned cosmetic lenses can get in the way though.

Does eye color affect iris recognition?

No. Scanners use near infrared light, which reveals iris texture even in very dark eyes.

Do identical twins have the same iris pattern?

No. Random events in the womb shape iris texture. So twins have different patterns, and even your own two eyes differ.

Is iris scanning harmful to your eyes?

Scanners use low power near infrared light. The FBI describes the risk from a single LED as remote. Well built devices follow eye safety standards.

Can iris recognition work after cataract surgery?

Often it does but not always. A 2019 study found match scores fell noticeably after surgery. You may need to enroll again.

Is iris recognition better than fingerprint or face recognition?

Each has strengths. The iris is highly distinctive and contactless. Fingerprints and faces are cheaper and more familiar.

Over to you

Iris recognition is fast and accurate and touch free. It’s also tied to something you can never change. That’s exactly why it sparks such strong feelings.

So where do you stand?

• Would you scan your iris to skip an airport queue or unlock your next device?
• Does it matter to you whether your iris data stays on your device or sits in a central database?
• Should private companies be allowed to collect iris scans to prove people are human online?

Share your view in the comments. There’s no single right answer here. There are only the trade offs you’re comfortable with.

Sources

1. How Iris Recognition Works: John Daugman, University of Cambridge
2. History of Iris Recognition: John Daugman, University of Cambridge
3. Iris Recognition: FBI Biometric Center of Excellence
4. IREX 10 report card: neurotechnology_020: NIST (August 2026)
5. IREX 10 evaluation results: Neurotechnology (May 2026)
6. About Optic ID advanced technology: Apple Support
7. Iris Recognition prevents entry of 351,318 deportees: Abu Dhabi Police (February 2013)
8. Aadhaar: Wikipedia
9. Aadhaar enrollment with iris scan if fingerprints unavailable: Outlook Business (December 2023)
10. CLEAR: Orange County Airport
11. Iris vs. retina biometrics: yes, they really are different: SecureIDNews
12. Samsung Galaxy S8 iris scanner hacked: The Hacker News (May 2017)
13. LivDet-Iris 2015: Iris Liveness Detection: NASK
14. New research raises questions about iris recognition systems: University of Notre Dame (July 2012)
15. Biometric aging: iris recognition: MITRE (2014)
16. Cataract surgery affects the discriminative capacity of iris pattern recognition: Scientific Reports (July 2019)
17. World Foundation scales proof of human: HackerNoon (July 2026)