Spectral Analysis of Human Face Expressions and Behavior Using Wavelet Transforms

Abstract

Face expressions communicate both the emotional state and behavioural intentions of a person. Behavioural tendencies of approach or avoidance of perceiver are also a function of the face expressions. Wavelet transforms extract information from many different kinds of signals due to its ability to simultaneously provide spatial and frequency representation of a signal. By discrete wavelet transform, the image is divided into four sub-bands using low pass and high pass filters. Using decimation process, the total size of the transformed image is the same as the original size.  Six types of excited face expressions namely angrily disgusted, angrily surprised, disgustedly surprised, happily disgusted, happily surprised and sadly disgusted are considered as test images. Haar wavelet transform with level-1 is used to find out scaling coefficients of the images having different facial emotional expressions. The scaling coefficients represent the trend of the image that occupies the upper left quadrant of the transformed image. The discrete wavelet transforms of images provide an accurate technique for extracting the face expression and expecting the behaviour of poser as well as perceiver.