Application of the Fourier-wavelet transform to moving images in an interview scene

Abstract

A new computer-aided method, based on Saito's Fourier-wavelet theory, is used to extract the facial expressions that appear and disappear in less than one-fifth of a second, i.e. expressions that reveal the true emotions of a human being. The new method consists of two major steps. The first step is the application of the Fourier transform to reduce the spatial phase differences between the reference and test images. The second is the application of the wavelet transform to extract the local discontinuous differences. This method is applied to the moving images in the interview scene. Using this method, the difference between the two facial expressions has been extracted, even though both images differ in their size and reference position.