Frederick Parke has written several seminal works on computerised facial imaging and animation [1,3]. In his Ph.D. dissertation A Parametric Model for Human Faces [3], he defined a facial model which is rendered as a mesh of Gouraud-shaded polygons from a set of real-valued parameters. Parke's model allowed both the expression and the conformation, or shape, of the face to be defined by parameters. In 1982, Parke revisited his model [3], refining it and referring to other work in the field. Andrew Marriott, of the Curtin University of Technology School of Computer Science, created a program for rendering faces defined in terms of Parke's model.
In [2], Ekman and Friesen defined a system (the Facial Action Coding System, or FACS) for encoding facial expressions in terms of a set of ``action units'' (AUs). Their system was designed for use by psychologists for noting down observed facial expressions, and defines six categories for complete expressions (happiness, sadness, anger, fear, disgust and surprise). Waters [4] based his facial animation system on FACS, mapping the action units to muscles in the face. Waters also renders the face using finer meshes of polygons, displaced smoothly so as to model skin elasticity.
Kalra et. al. [5] developed another technique for more realistically rendering facial images; they used rational free form deformations, a method of distorting curved surfaces by manipulating grid points surrounding them. The facial model used is similar to FACS, in that it consists of perceivable actions; it consists of 21 Minimum Perceptible Actions (MPAs), each represented by a real value.