Abstract
The process of intelligent interaction through brain machine interface
requires quick and accurate extraction of Electroencephalogram (EEG)
signals. However, the accuracy of signal classification varies with the
signal extraction location. Is there a universal rule to follow to
determine the optimal extraction location? This paper investigates the
possibility of a universal rule to determine optimal extraction location
through Welch, Support Vector Machine and Euclidean distance algorithms.
The motor imagery EEG signals of 40 subjects were extracted and the
classification correct rates of brain electrode signals in different
positions were analyzed using Welch and Support Vector Machine
algorithms. Then the electrodes were sorted according to the correct
rate, and finally three pairs of electrodes with the highest correct
rate were obtained. For comparison, this paper proposed another method
of searching for the optimal electrodes, namely the combination of Welch
and Euclidean distance algorithms. Ultimately, a similar conclusion was
drawn from the above two methods: T3/T4, F3/F4 and C3/C4 electrodes
usually have high classification accuracy. This result is helpful for
quick customization of optimal brain electrode placement.