Skip to main content
. 2019 Jan 26;19(3):522. doi: 10.3390/s19030522
Variables Considered.
Input:
  EEG Signals: the EEG Signals from which extracting the CSP features. These signals are of a structure such that:
   EEG Signals.x: the EEG signals as a [Ns × Nc × Nt] Matrix where
      Ns: number of EEG samples per sample
      Nc: number of channels (EEG electrodes)
      Nt: number of samples
     EEG Signals.y: a [1 × Nt] vector containing the class labels for each sample
     EEG Signals.s: the sampling frequency (in Hz)
  CSP Matrix: the CSP projection matrix
    nbFilterPairs: number of pairs of CSP filters to be used. The number of features extracted would be twice the value of this parameter. The filters selected are the one corresponding to the lowest and highest eigenvalues.
Onput:
  Features: the features extracted from the above mentioned EEG data set as a
[Nt × (nb Filter Pairs × 2 + 1)] matrix, with the class labels as the last column.