In the information era, information security becomes more important, the use of EEG as identification tool become more and more. In this paper, we used the subjects’ photos as stimulation. In order to obtain the Identification classifier of different subjects, we used AR model to convert the EEG signals from time domain into the frequency domain, used Fisher’ distance to extract the feature. Finally, we calculated the feature by using BP neural network. Through the analysis of the correct recognition rate, error recognition rate and false recognition rate, we achieved the purpose of using EEG as identification tool.