Expression recognition from time-sequential facial images by use of expression change model

In order to construct a better human interface, recognition of facial expressions by means of computer is an important technology. An approach is proposed to recognize the degree of facial expression change from time-sequential images. The facial features in an input image sequence are tracked by using labeled graph matching with weighted links. To represent the relationship between the motion of features and change of expression, we construct expression change models by using B-spline curves. By making a comparison between the trajectory of features and the expression change models, the facial expression in the input image sequence can be recognized. Not only the category, but also the degree of facial expression change can be determined. Furthermore, the obtained expressional information is then fed back to guide the tracking in the next frame