A Study Of Emotion Recognition Based On Physiological Signals

With the development of computer technology and pervasive computing technology, emotional expression and perception is becoming increasingly important as an adjunct to interpersonal interaction. In recent years, more and more people are concerned about how to make computer have the ability to process and express emotions. It requires that a computer or robot can replace and auxiliary humans engage in more extensive and more complex work. At the same time, it is also requires to have more harmonious and friendly human machine Interface to replace, compensation and strengthen people’s presence capability, the function of thinking and behavioral functions in an increasing number of aspects. This necessarily requires a computer and the robot have a stronger capabilities of emotional expression, emotion recognition and emotional understanding.In the introduction section, this paper briefly describes the background and significance of emotion recognition based on physiological signals.Then, the paper summarizes current domestic and international research status of emotion recognition based on physiological signals. It also has a brief introduction of the recognition process based on physiological signals and related technologies. Because of the characteristics of physiological signals,the paper introduces relevant knowledge of wavelet de-noising for Signals in the next section. Firstly, it introduces several methods of wavelet denoising, including modulus maxima de-noising, the denoising based on the correlation between the scale transformations, Thresholdingdenoising and so on. Then,it introduces the basic theory of AdaBoost algorithm. In this algorithm, each sample is given a weight which represents the probability the sample was selected to training subset. In an iterative process, if one sample is properly classified before, then its weight will increase, whereas the weight is reduced. In this way, the algorithm focus on those samples that is difficult to distinguish,and thus improve the classification accuracy of difficult samples. This article uses AdaBoost algorithm to classify emotions of physiological signals.The paper also uses Fisher as AdaBoost base classifier. Experimental results show the feasibility of applicable AdaBoost algorithm to emotion recognition.