Tracking Human Faces in Real-Time,

Abstract
Many applications in human computer interaction (HCI) require tracking a human face. In this report, we address two important issues for tracking human faces in real-time: what to track and how to track. We present a stochastic model to characterize skin-colors of human faces. The information provided by the model is sufficient for tracking a human face in a various poses and views. The model can be adapted in real time for different people and different lighting conditions while a person is moving. We then present a model-based approach to implement a real-time face tracker. The system has achieved a rate of up to 30+ frames/second using an HP-9000 workstation with a framegrabber and a Canon VC-Cl camera. It can track a person's face while the person moves freely (e.g., walks, jumps, sits down and stands up) in a room. Three types of models have been employed to track human faces. In addition to the skin-color model used to register the face, a motion model is used to estimate image motion and to predict search window; and a camera model is used to predict and to compensate for camera motion (panning, tilting, and zooming). The system can be applied to tele-conferencing and many human-computer interactive applications such as lip-reading and gaze tracking. The principle in developing this system can be extended to other tracking problems such as tracking the human hand for gesture recognition.

This publication has 0 references indexed in Scilit: