A Method of Emotion Recognition Based on ECG Signal

Emotion recognition from electrocardiography (ECG) signal has become an important research topic in the field of affective computing. In the current work, ECG signals from multiple subjects were collected when film clips shown to them, and then feature sets were extracted from precise location of P-QRS-T wave by continuous wavelet transform (CWT). Hybrid particle swarm optimization (HPSO) was utilized for feature selection, whose discrimination criteria was the correct rate of fisher classifier and the number of features selected. For recognizing two emotions of joy and sadness, effective features and better recognition rate were obviously obtained. Experimental results indicate that the features that acquired from experimental simulation can represent the changes of emotions, HPSO and fisher classifier are effective ways for emotion recognition.