Emotion Recognition Based On Sample Entropy Of EEG

Emotion recognition is a hot topic of current research, which belongs to the field of artificial intelligence. Human emotion and cognition research is the advanced stage of artificial intelligence. It is greatly significance to study how human brain deals with the emotional state of us for exploring the operation mechanism of human brain. Emotion recognition played an important role in our daily life. As a result, lots of ways to study human emotions comes into being. EEG is one of the primary means to study human emotion.Emotion recognition based on EEG is used widely. The main contents of this paper are as follows:(1) We proposed an emotion recognition method based on sample entropy of EEG. After artifact removal and filtering, and through screening notable channel by K-S test, the emotional classification feature vector was formed using EEG signals, and then used S VM-Weight algorithm for classification.(2) Firstly, we designed a psychological experiment which the stimulus was based on pictures, and then used EEG recorder system provided by BP to collect EEG signals. Secondly, we used the preprocessing software provided by BP to preprocess the EEG signals, and then extracted the β-wave of the preprocessed EEG signals. Lastly, we recognized the positive and negative emotional states of the (3-wave EEG signals based on the method we proposed.(3) Firstly, we screened the videos which the behavioral experiments noting was in accordance with subjects noting from preprocessed emotional EEG signals provided by Deap set website. Secondly, we extracted the β-wave of the preprocessed EEG signals. Lastly, we recognized emotional states of different degree of arousal and valence of the β-wave EEG signals based on the method we proposed.(4) The emotion recognition accuracy and feature extraction efficiency of Sample entropy was compared with it of the other three feature extraction methods (approximate entropy, LZC complexity and Hurst exponent), the results illustrate that compared to the other three features, sample entropy is more suitable for EEG feature extraction and emotion recognition.In summary, the results of this paper adequately showed that using sample entropy as EEG features for emotion recognition has a reasonable recognition effect, also confirmed the possibility of β-wave EEG signals for emotion recognition, and also found the brain areas related to emotion recognition activities. This method is expected to have a good effect in BCI, health care systems and other intelligent applications fields.