PROCEDURE
1. Capture
B/W video image is captured:
- Inner and outer boundaries at limbus and pupil are captured
- Contraction and dilatation of pupil are accounted for
- Artifacts (eye lashes, light reflexes, etc.) in the visualization area of the iris are reduced
- Image is oriented in a single plane
2. Iris code is generated
Iris features are converted into the 512 byte iris code. (256 bytes for the features, 256 bytes
for control)
3. Code is stored
IrisCode® is stored in a database or on another
token in encrypted format.
4. Recognition
IrisCode® is compared against all IrisCode® records in a database or on another token.
5. Result
Decision: "match" or "no-match".
"The decision for "match" or "no-match" is based
on the statistical value of the Hamming Distance (HD).
General rules:
- if more than 68% of all bits match, it's an Accept.
(more than 68% = HD < 0.32 for up to 18 million users)
- if less than 68% of all bits match, it's a Reject.
(less than 68% = HD > 0.32 for up to 18 million users)
- 0.28 is sufficient to securely identify 84 billion people.
- Inner and outer boundaries at limbus and pupil are captured
- Contraction and dilatation of pupil are accounted for
- Artifacts (eye lashes, light reflexes, etc.) in the visualization area of the iris are reduced
- Image is oriented in a single plane
"The decision for "match" or "no-match" is based on the statistical value of the Hamming Distance (HD).
(more than 68% = HD < 0.32 for up to 18 million users)
(less than 68% = HD > 0.32 for up to 18 million users)


