Welcome To Our Search Page

Real Time Movement Imagery Classification based on Reverse filtration process

    Dept. of ECE, Sagar Institute of Science and Technology, Bhopal, Madhya Pradesh, India


Abstract : Recent advancement in biomedical research work, increases the demand of real time algorithm for Electroencephalogram (EEG) signal processing. In this paper the authors highlighted Mel Frequency Cepstral Coefficient (MFCC) based reveres filtration algorithm for online (i.e. real time) as well as offline motor imagery classification for Brain Computer Interface (BCI). The proposed MFCC based algorithm is very first and robust unsupervised algorithm for offline and online movement imagery classification. The difference of the different Hjorth parameters is taken from the cepstral coefficients has been taken as feature for movement imagery classification. The classification accuracy and mutual information has been taken as BCI evolution parameters. The extracted features has been classified using linear classifiers for offline comparison. The offline processing algorithm has been applied in nine subject data of BCI competition IV dataset 2b. The online processing algorithm has been applied in the self recorded single subject data set. The real time processing algorithm has been compared through the performance in training and testing dataset.

Keyword : Brain-computer interface (BCI), Classification Accuracy, Mutual Information, Electroencephalogram (EEG), Mel Frequency Cepstral Coefficient (MFCC) and Movement Imagery (MI)