Analysis of facial EMG signal for emotion recognition using wavelet packet transform and SVM

Abstract

Emotion recognition has been improved recently and effectively used in medical and diagnostic areas. Automatic recognition of facial expressions is an important application in human–computer interface (HCI). This paper proposed techniques for recognizing three different facial expressions such as happiness, anger, and disgust. Facial signals were recorded using two-channel wireless data acquisition system. Recorded facial EMG signals from zygomatic and corrugator face muscles were set up in four steps: Feature extraction, features selection, classification, and emotion recognition. The features have been extracted using wavelet packet transform method and feed to support vector machine for the classification of three different facial emotions. Finally, the proposed methodology gives classification accuracy 91.66% on 12 subjects.