Dynamic display of CG facial expressions for active human interface

In order to develop an active human interface that realizes heart-to-heart communication between intelligent machine and human being, we´ve been undertaking the investigation of method for improving the sensibility or “KANSEI” communication by using facial expressions. To express 6 basic facial expressions dynamically by computer graphics (CG) in a similar manner to that in human being, we develop the method of dynamic control of the movement of grid points of wire frame model of human face by applying a neural network. Using the experimental data obtained from dynamic facial expressions, time-dependent change in the magnitude of AUs needed for expressing each of 6 basic facial expressions is formulated by the neural net mapping technique. And the face components in the wire frame model are dynamically forced to move through the time-dependent AUs, resulting in success to display the dynamic facial expression of 6 basic emotions at the rate of 30 pictures per second