An Identification System Based on Portable EEG Acquisition Equipment

EEG-based identification system can effectively improve the security in the existing biometric systems because of its uniqueness, stealthiness, non-imitability and receipt-freeness. However, owing to the drawbacks in data acquisition equipments and signal feature extraction methods, the development of EEG-based identification system remains in the experimental stage. In this paper, portable acquisition equipment was used for data collection. After evaluating the signals, the AR model parameters were extracted from the valid datum as the identification features. Combined with these features, an identification system was realized based on three pattern recognition methods including BP neural network, support vector machines(SVM) and linear discriminant analysis(LDA). The conclusions indicate that: 1. The data quality can be effectively improved when the subject is fully relaxed and keep his/her eyes closed during data collection. 2. The AR model parameters of EEG signals can be used for identification. 3. The efficiency of SVM is the highest and the identification rate is up to 87%.