Extraction and application of dynamic pupillometry features for biometric authentication

• A multimodal biometric system based on the Pupillary Light Reflex is proposed.
• A pupillometer device has been mounted and authentication tests have been conducted.
• Human reflex as biometric trait can avoid spoofing by a fake organ.
• The dynamic features alone are discriminant, but not enough for authentication.
• Fusion of dynamic features with iris pattern leads to a better performance.

Iris recognition is considered one of the most secure biometric recognition methods. However, even this biometric trait can be frauded in some cases. In other cases, the quality of the captured image can affect the system recognition performance. This paper proposes a multimodal biometric authentication system that combines the use of dynamic features from the Pupillary Light Reflex (PLR) and the static features from the iris pattern for a better performance. A dynamic pupillometer device has been developed and a prototype system for features extraction and classification has been implemented. A pupillometric database has been created using data from 90 volunteers, and tests of the biometric system have lead to experimental of 0% EER.