Research on the Recognition of Motor Imaging EEG Signals Based on the PSD and the SVM

In the study of brain-computer interface, the EEG signal recognition classification method of motor imaging is one of the hot spots in today's research. In this study, we use frequency domain analysis based on the power spectrum density function to extract features from motor imagery EEG signals for two different motor imagery movements. We then combine with SVM to classify the features because BCI systems currently still have a low recognition rate accuracy. This study finds that employing the power spectrum in conjunction with SVM improves EEG signal classification when compared to alternative classifiers, such as random forest and convolutional neural networks. It can offer algorithmic assistance for the widespread use of BCI systems in everyday life.