Skip to main content

Table 2 Rizon’s channel selection algorithm for emotion classification [68]

From: A review of channel selection algorithms for EEG signal processing

1. The raw EEG signals from five subjects over five discrete emotions are collected using 63 electrodes which are placed through standard International 10/20 system on the scalp.

2. The signals are pre-processed by mean removal and variance normalization.

3. The signals are filtered using 5th order band-pass filter at a cutoff frequency of 0.05 Hz – 45 Hz.

4. The signals are divided into five different EEG frequency bands using 5th order Butterworth filter.

5. Alpha band is used for channel selection.

6. From the 63-channel EEG signals, only 28 homogeneous channel-pairs are separated out for calculating the asymmetric ratio for channel selection.

7. For each subject, the values of AVR and AAR are calculated for all the 28 homogeneous pairs of electrodes using the equations (6) and (7).

8. If all values of AR are positive or negative for five emotions this leads to a rank of 5, for four emotions this leads to a rank of 4, etc.

9. The channel pairs of higher ranks of AAR and APR are sorted as the dominant channels for emotion recognition.