Abstract
Electroencephalography (EEG) signals have been utilized in various biometric research. Although the biometric of EEG given its advantageous in anti-counterfeiting, the forecasts accuracy and efficiency of the current systems are still unsatisfactory in practical application. In this paper, a new approach to extract features for EEG based biometrics is proposed, which can achieve high accuracy and efficiency. The proposed method uses the High Order Statistic and different entropy as features. Evaluation of time and frequency-dependent features have been examined to investigate their distinctive ability. In this paper, the method used to assesses participants under normal emotional states without motor movements. The experimental results show that this new approach can reduce the heavy computational load and ensures a relatively high classification accuracy in recognizing the users. The average accuracy performance achieves to 95.7% which significantly improve than the previous study.