Features Extraction of Different Emotional based on Brain Rhythm

Emotions are an essential part of human physiological performance. Therefore, the study of emotion recognition is extremely relevant in both practical applications and theoretical research. Based on the electroencephalogram (EEG) data we collected, this paper explored the differences from five different characteristics among the coefficient of variation (CV) , power spectral density (PSD) , differential entropy (DE) , approximate entropy (AE) , and permutation entropy (PE) of three emotions at different rhythms, analyzed the activity of each brain region under different emotions. Finally, this paper also compares the classification accuracy of single features and combined features. The experimental results reveal that emotions are more strongly correlated with Beta and Gamma bands; temporal lobe regions are most active in emotion-related brain activity; and the accuracy of combined features is higher than the accuracy of single feature classification. The experiment conclusively demonstrates that effective feature extraction and selection have far-reaching implications for emotion recognition.