The research of EMG signal in emotion recognition based on TS and SBS algorithm

This study is on the emotion recognition based on electromyography (EMG) signal. EMG signals from multiple subjects were collected when film clips were shown to them, next all of the data were de-noised and original features were extracted with wavelet transform method, then feature selection was done using the Tabu search (TS) algorithm combined with the Sequential Backward Selection (SBS), at last in recognition of six emotions (happiness, surprise, disgust, sadness, grief, anger and fear), better recognition rate were obviously obtained based on selected feature subset. Results show that through the optimization of TS algorithm with SBS, the dimension of feature subset was reduced, and the emotion recognition of physiological signal also achieved good effects