Personal Authentication with Eye Movement Features During PIN Input

In recent years, personal authentication is used in various situations. Among them, the personal authentication using biometric information is becoming widespread. Conventional biometrics have a spoofing problem. Personal authentication using eye movement is considered to be more difficult to spoof. Biometrics research using eye movement has been conducted in the case where the eye movement range is relatively wide. In this study, we aim at development of the personal authentication method using eye movement in the case of narrow range of eye movement such as smartphone and ATM operation. In this paper, we investigated effective features of eye movement in the case where the range of eye movement is narrow. The authentication accuracy was evaluated using equal error rate (EER). In addition, we examined whether the authentication accuracy could be improved by score level fusion. From the experimental results, it was found that complex eye movement (CEM) had the best authentication accuracy when the range of eye movement was narrow, and the authentication accuracy was improved by score level fusion.