No abstract

Classification of electroencephalogram (EEG) data for different motor imagery (MI) tasks is a major concern in the brain-computer interface (BCI) applications. In this paper, an efficient feature extraction scheme is proposed based on the discrete wavelet transform (DWT) of the EEG signal. The EEG data of each channel is windowed into several frames and DWT is performed on each frame of data. Considering only the approximate DWT coefficients, a set of statistical features are extracted, namely wavelet domain energy, entropy, variance, and maximum. In order to reduce the dimension of the proposed feature vector, which is composed of average statistical feature values of all channels, principal component analysis (PCA) is employed. For the purpose of classification, k nearest neighbor (KNN) classifier is employed. Proposed classification scheme not only offers significant reduction in feature dimensionality but also provides satisfactory classification accuracy. For the purpose of performance analysis, publicly available MI dataset IVa of BCI Competition-III is used and a very satisfactory performance is obtained in classifying the MI data in two classes, namely right hand and right foot MI tasks.

No abstract

No abstract

In this paper, an efficient scheme is proposed for classification of electroencephalogram (EEG) data in to different motor imagery (MI) tasks based on empirical mode decomposition (EMD). The EEG data recorded from each channel is first decomposed into a set of intrinsic mode functions (IMFs) by using the EMD analysis. In view of extracting discriminative information from the EEG signals corresponding to different MI tasks, we propose to utilize the entropy of bandlimited IMF. Instead of considering all IMFs or first few IMFs, in the proposed method only the first IMF is chosen because of its low variance. In order to reduce the dimension of the feature vector consisting of entropy values from all channels, principal component analysis is performed. For the purpose of classification, train and test datasets are formed as per leave one out cross validation scheme and then linear discriminant analysis (LDA) is carried out. Simulation is performed on publicly available MI dataset IVa of BCI Competition-III to classify the MI data in to two classes, namely right hand and right foot MI tasks. It is observed that the proposed classification scheme not only offers significant reduction in feature dimensionality but also provides satisfactory classification accuracy.