Research And Analysis On Correlation Between Physiological Signals And Emotions

Around the world,there are few researches on the correlations between physiological signals and emotions at present,and it's helpful to improve the accuracy of emotion recognition while knowing the correlations between them.This paper researched and analyzed the correlations between physiological signals and emotions.In this paper,the emotional physiological data from the databases of Augsburg University(Germany)and the MPED of Southeast University(China)were used,and the data of electrocardiograph signal,respiratory signal and skin conductance signal under the emotions of joy,anger and sadness were taken as the research objects.Firstly,extracted chaotic features of the sample,included the maximum Lyapunov index,fractal dimension and complexity.Secondly,based on the data of Augsburg University and MPED,Grey Relational Analysis and Multinomial Logistics Regression were used to complete correlation research and result analysis respectively.In order to improve the accuracy of emotion recognition,this paper optimized the emotion recognition algorithm,through using Bagging to optimize the CHAID decision tree.On this basis,emotions were carried out on the electrocardiograph,respiratory and skin conductance from the adopted databases by Bagging-CHAID decision tree algorithm,and identification results and correlation results were analyzed synthetically.The results show that:1)The correlations are the same while using the same assoication analysis method based on different databases(Augsburg University database and MPED).2)Compared with the CHAID decision tree before optimization,the optimized CHAID decision tree(Bagging-CHAID decision tree)has better sentiment classification performance.It's also learned that the recognition accuracy of joy,anger and sadness through respiratory is the highest,followed by skin conductance and electrocardiograph.3)Based on the same database,using different correlation analysis methods(Grey Relational Analysis and Multinomial Logistics Regression)can obtain different correlations.However,the experiment shows that the result obtained by Grey Relational Analysis are closer to the objective result.4)The correlations between physiological signals and emotions obtained by Grey Relational Analysis are that respiratory has the largest correlation with joy,anger and sadness,followed by electrodermatogram,and electrocardiogram has the least correlation with these three emotions.