Eye-blink based Personal Authentication Using Time-series Directional Features and Waveform Features
Keisuke Takano, Hironobu Takano · 2019
In this study, we propose the personal authentication method using characteristics of eye blink and investigate effective features for authentication. The time-series gradient directional features and the waveform features extracted from the time-series gradient intensities are adopted as the features for recognition. The degree of similarity between registration and recognition features is obtained by the dynamic time warping (DTW) and Euclidean distance. From the experimental results, individual differences appeared in features obtained in the iris peripheral region and during opening eyes.