| With the continuous development of human-computer interaction technology,affective computing(AC)has gradually become an emerging field of emotion research,and emotion recognition is an indispensable part of emotional computing.In recent years,experts in artificial intelligence have devoted themselves to adding emotional elements to artificial intelligence,and the medical community is also working hard to solve the problems of mental illness such as depression and autism.Therefore,research on emotion is very meaningful and valuable.The data studied in this thesis comes from 20 healthy college students.The research points mainly include the following aspects:(1)The feature extraction method of EEG signals was studied.A dual-tree complex wavelet transform is proposed to decompose and reconstruct the EEG signal algorithm.Compared with the traditional wavelet transform,the algorithm has good anti-aliasing and translation invariance.The algorithm extracts the phase and energy information of EEG signals under different emotions.Using non-linear dynamic analysis methods,the classification effects of the three methods of approximate entropy,Hurst exponent,and fractal dimension are compared.(2)The method of pattern classification of EEG signals was studied.The support vector machine(SVM)classification method is mainly used to study the problem of finding the optimal parameters by improving the traditional grid search method when the radial basis kernel function is selected.The simulation results show that the improved algorithm is obviously Improve the efficiency of finding the optimal parameters.(3)Emotion recognition of EEG signals was studied.The feature extracted by the dual-tree complex wavelet transform is combined with the nonlinear dynamic feature fractal dimension.By recognizing the three emotions of calmness,happiness,and sadness,the results show that the proposed algorithm has a higher recognition rate than previous studies. |
