Research Of Emotion Recognition System Based On EEG-fNIRS

Emotion comes from the psychological level,which refers to the attitude experience of human beings towards external things.In a broad sense,emotion can be quantified as positive emotion and negative emotion.At present,with the rapid advancement of China's Industry 4.0 and China Made 2025 plans,emotion recognition technology is widely used in the core fields of intelligent robot,industrial internet and artificial intelligence.In addition,the huge pressure of work and life in today's society has caused a sharp increase in the number of patients with mental illnesses such as depression and anxiety.Therefore,the diagnosis and treatment of psychological diseases characterized by negative emotions has become the focus of clinical work.There are three methods of emotion recognition.The first one is based on facial expression or voice tone to identify emotions.The second one is based on peripheral physiological signals such as ECG,EMG and pulse to identify emotions.The third method is based on central nervous system signals and uses non-invasive brain function detection technology to identify emotions.In the research of emotion recognition based on central nervous system signal,the method of emotion recognition based on electroencephalogram(EEG)is often used,which can achieve emotion classification by extracting multi-dimensional EEG features and complex algorithm.Its disadvantage is that the representation of brain activity is not intuitive enough,and it lacks the analysis of emotion inducing effect.In recent years,the functional near-infrared spectroscopy technology fNIRS has become an emerging brain function detection method.Although its application to emotion recognition is still in its infancy,it can not achieve a good recognition effect,but fNIRS can intuitively characterize changes in brain activity.There is an advantage in analyzing the emotioninduced effects,which can make up for the lack of EEG-based emotion recognition methods.In addition,there is no mutual interference between EEG and fNIRS,which can realize simultaneous measurement and can be portable.Therefore,this paper studies an emotion recognition system based on EEG and fNIRS,analyzes the evoking effects of emotional stimulation experiments,and realizes the classification of positive and negative emotions.The main work of this paper includes,firstly,the 16 channel EEG signal and 8 Channel EEG signal acquisition system are designed;secondly,the modified Lambert Beer law is used to calculate the change of cerebral blood oxygen,and the statistical characteristics and change trend characteristics of cerebral blood oxygen data are extracted;thirdly,extract the time domain,frequency domain,time-frequency domain,and space domain characteristics of EEG,use the principal component analysis method to reduce the EEG characteristics,and use support vector machines to build a sentiment classification model;finally,according to the difference between the baseline data and the stimulation data of cerebral blood oxygen characteristics,analyze whether the emotional stimulation experiment successfully induced the subjects to produce emotional changes,select the experimental data successfully induced,and then use the EEG classification model to classify the positive and negative emotions.This paper has completed the design of a simultaneous acquisition system of EEG and fNIRS signals including a hardware detection platform and a host computer processing platform.Through emotional stimulation experiments,it has been proved that the extracted cerebral blood oxygen characteristics can effectively judge the induced effect of emotional stimulation experiment,combined with cerebral blood oxygen Features and EEG emotion classification model can effectively realize emotion classification and improve classification accuracy.