Multi-cue facial feature detection and tracking under various illuminations

An efficient and robust facial feature detection and tracking system under various illuminations is presented in this paper. A mixture of global and local illumination balance methods is proposed to compensate for the variations in illumination and to accentuate facial feature details. The system is capable of locating a human face automatically. Six facial feature points (pupils, nostrils and mouth corners) are detected and tracked using multiple cues including facial feature intensity and its probability distribution, geometric characteristics and motion information. In addition, to improve the robustness of the tracking system, a simple facial feature model is used to estimate the relative face poses. This system has the advantage of automatically detecting the facial features and recovering the features lost during the tracking process. Encouraging results have been obtained using the proposed system.