Automatic Person Authentication Using Fewer Channel EEG Motor Imagery

A biometric person authentication system using brain waves or Electroencephalogram (EEG) signals recorded using a minimum number of channels ranging from 2 to 6 is presented. The task for EEG recording consists of simple motor imagery movements that the subject has to imagine. The system uses an effective time-frequency based feature extraction method using the short-time Fourier transform (STFT) or spectrogram. Energy, variance, and skewness features are computed on the spectrogram. These features are used to train a support vector machine and neural network classifier. The classifiers are tested for person authentication with testing data using cross-validation. Results using a different number of channels with optimum features are presented.