Analysis of human affect recognition Using hybrid features with Multidimensional binary Tree svm

Abstract

Affect identification is a methodology to understand the type of newlineemotion a person expresses and to develop an efficient automated system to newlineidentify emotions in human beings is a primary challenging task. The building newlineof the structure is centred on the interrelationship among the magnitude of newlineemotion inducement by video, audio content, and human response to it. The newlineaffective changes in human beings bring changes in facial expression, speech, newlineiris, and also in physiological signals. This change helps to improve an newlineautomatic affect determination method, widely can be used in many newlineapplications. newlineThe initial step in the development of affect identification is to newlinemodel the affect state. Psychological researchers have categorized affect state newlineinto discrete form. One of the models classifies the affect state on a newlineone-dimensional scale as six basic types surprise, happiness, sadness, anger, newlinefear, and disgust. Another methodology split the affect state discretely based newlineon daily life. The main drawback of using a discrete model for emotion is that newlinemore circumstances and realistic analyses are required to describe feelings, newlinechallenging to develop an automatic affect recognition system. newlineThe two-dimensional scale modelling is used to overcome the newlinedrawback in a one-dimensional model. There are two major types of newlinetwo-dimensional models; one is the emotional wheel model suggested by newlinePlutchik. The other model proposed by Russell is the valence-arousal model. newlineThe affect states well defined in the valence-arousal model; hence this model newlineis commonly used. In this model, valence denotes the positive or negative newlinefeelings along the x-axis and arousal define the affect state on the scale of newlinebored or excited feelings along the y-axis newline