This paper proposes a new method of transforming facial expression based on
facial sketch image acid multiscale edges. Multiscale edges, which are def
ined as local maxima of high-frequency signal of facial images, are general
ly located on the edges of facial parts. Facial sketch images defined as li
ne drawings of edges of facial parts are extracted by morphological process
ing, the generalized symmetry operator the rectangle filter, and a geometri
cal template. Therefore, it is possible to transform the binary sketch imag
e into gray-tone multiscale edge images which are called characterized sket
ch images. The expression of the characterized facial image can be transfor
med easily based on FAGS (Facial Action Coding System). By reconstruction b
ased on Mallat's method of reconstruction of multiscale edges, facial image
s with different expressions can be obtained easily. (C) 2001 Scripta Techn
ica.