Research And Exploration Of Identity Recognition Based On Visual Evoked EEG

Today,biometric identification system has become the mainstream key unlocking way,but the current existence of biometric identification system such as fingerprint,facial recognition,voice and other biometric identification system,can be broken by criminals,opportunity.In this paper,the excitation,characteristics and classification of EEG signals are studied,and the EEG signals are used as identification keys to explore the new biosafety identification system.Firstly,the generation method of visual induced EEG signal is studied.In view of the single excitation source and poor effect in existing experiments,selected experimental stimulus materials were used to randomly display the excitation materials and generate P300 signals at fixed time intervals to improve the experimental paradigm,refine the experimental operation and construct the EEG data set for identification.Secondly,the EEG signal of identity recognition is extracted by feature extraction.Artificial,ICA and average non-target EEG signals were used to preprocess EEG signals.FPE and AIC criteria were studied,and the optimal AR order was selected to construct the autoregressive model of EEG signals for identification.The characteristics of AR coefficients were solved by Burg algorithm.The wavelet decomposition of EEG signals for identification is studied.The EEG signals are decomposed by Mallat algorithm,and the detail components of ? and ? bands related to attention are extracted for digital feature analysis,and the wavelet decomposition features are extracted.By comparing single model feature and combination feature,the EEG signal feature of identification is determined to be the combination feature of AR coefficient feature and wavelet decomposition feature.Finally,the classification and processing of EEG signals in identity recognition are studied.The number of channels decreases in pairs to compare the performance of multi-channel and single-channel EEG signals,and the best CZ channel is selected as the main research channel,which is conducive to the promotion of portable EEG devices.The experimental results show that the 13-dimensional AR coefficient features and the 6-dimensional wavelet decomposition features selected in this paper can achieve 97.2% recognition accuracy under the use of SVM classifier after feature combination.The stability of the identity recognition system was analyzed,and a variety of features,classifier comparison and personal characteristics comparison of different subjects were selected to explore the construction of a reliable identity recognition system.