For optimal matching of the iris images, iris recognition algorithms require iris images to be captured at or near orthogonal plane - the subject directly looking at the camera. In many instances, iris images may be captured off-axis. This may be due to:
- Non cooperative/unwilling subject
- Distracted subject, looking elsewhere
- Iris Scanner is mounted too high or too low
- Iris images captured in a non-voluntary/covert setting
- Untrained operator
In a controlled capture scenarios, with a trained operator or an intelligent camera can detect severe off-axis iris images and can initiate a re-take of the image. But in many cases, re acquiring the iris image may not be an option. The new breed of iris recognition algorithms to some extent, can extrapolate information that was distorted or skewed due to off-axis capture. But effectiveness of the off-axis correction in general, depends on the severity of the off-axis angle, uniformity of the illumination and focus/sharpness across the iris region. At severe off-axis angles, We noticed lager structures in the iris can cast shadows that can interfere with quality of the iris signatures. Above is an example of our Iris Authenication algorithms performing on an off-axis iris image.
NOTE: The iris regions in these images are altered/blurred to protect the privacy.