International Journal of Engineering and Technology

This paper proposes a unique feature extraction method based on linear convolutive mixing model of the human brain for the real time application. Proposed method is very useful for electroencephalogram (EEG) signal based real time brain computer interface (BCI). The raw EEG signals are subjected to band pass filter to select the band of interest. The filtered EEG signals are subjected to the proposed feature extraction method. The proposed feature extraction method considers the Multivariate Gaussian Probability Density Function (MVGPDF) of Cepstral Coefficients (CC). The MVNGPDF is applied to generate the probabilistic features over cepstral coefficients. Further, the extracted feature is subjected to the conventional linear classifiers like Naive Bayes and linear discriminant analysis classifier to decide its belongings. The performance of the proposed feature extraction method is compared in terms of the classification accuracy and mutual information.