👁️ The iris (the colored part of the eye) has a pattern that is unique to each person, even identical twins. That’s why scanning the iris is a super precise way to verify someone’s identity.
🧪 How does this process work?
Iris capture: The Orb takes a picture of your eyes.
Gabor wavelet processing: An algorithm transforms that image into an "iris code", something like a fingerprint, but of your eye.
Uniqueness verification: That code is compared with all existing codes to ensure there are no duplicates.
Registration in the system: If your code does not match anyone else’s, it is confirmed that you are a unique person and your identity is recorded.
🧮 What can go wrong?
There are two types of possible errors:
False Match Rate (FMR): The system thinks you are someone else. (Severe error).
No Match Rate (FNMR): The system does not recognize you even though you are already registered. (Annoying, but less severe).
⚖️ What rules do they use to avoid errors?
They use two ways to combine the information from both eyes to enhance accuracy:
OR Rule: If one of the eyes matches, you are accepted. (Easier to register, but less secure).
AND Rule: Both eyes must match. (Safer, but more demanding).
💡For millions (or billions) of people, the AND rule is more effective in the long run because it reduces severe security errors.
📈 How is it performing today?
At first, the results were not that good in real environments (on the street, with changing light, etc.).
But with improvements in hardware and the use of artificial intelligence, performance is now very high even outside the lab.
This means that the system can scan your iris in real life and still accurately know that it is you and that there is no other like you.
🚀 So what’s next?
Even smarter algorithms are being developed, based on deep learning, that do not rely on fixed rules but learn on their own with millions of data.
This could bring iris recognition to levels of accuracy never seen before.
