Study On Emotion Recognition Methods Ofanxiety State Based On Physiological Signals

Anxiety is a kind of complex emotional reaction when someonefeeling threatened. Usually this kind of emotion is generated because theoriginal task is too challenging to be finished. It is a kind of combinationof reflection, including the physiological reactions, personalcharacteristics and social effect of human. The physiological signals are akind of bio-electricity signals or bio-impedance signals, which areregulated by human automatic nervous system and endocrine system, andcannot be easily influenced by the society features, which makes thismethod more real and objective.In this paper, physiological signals of subjects were collected duringthe emotion elicitation experiments. Based on the multi-physiologicalsignals generated by the subjects in emotions, such as heart rate andrespiration rate, the Relief algorithm was adopted to optimize the featuresselection from the features of multi-physiological signals, and combinedwith k-Nearest Neighbor (kNN) and least squares support vector machineclassifiers (LS-SVM) to classify the anxiety state and the normal state.The results showed that the classification accuracy of the LS-SVMalgorithm was better than the kNN algorithm when both were combinedwith Relief algorithm. The most useful characters were summarized andthe model of emotion recognition of anxiety state based on physiologicalsignals was built up. The model of emotion recognition of anxiety state based onphysiological signals is significant to the assessment of emotion andcontrol in the psychological area.