SINOPSIS

A new computer-aided method of extracting facial features has been proposed. The conventional method has drawbacks because of the analysis of extracted expressions through researcher's eyes. The new method based on the Saito's Fourier—wavelet theory consists of two steps. The first step is the application of the Fourier transform to correction of positional differences between the reference and test images, and the second step is the application of the wavelet transform to extraction of minute differences between the two images. Using this method, the difference between a normal and a smiling face has been extracted, even though tire two pictures differ both in size and center position.