Skin-Color Based Face Tracking in Real-Time

A robust real time face tracking algorithm for tele conference, tele education, surveillance and monitoring was presented. The face motion could be estimated by using face skin color model integrated with feature based object recognition technique and extended Kalman filter. The system worked at video rate, that is, 25 Hz for NTSC images and 30 Hz for PAL images. The experimental testing showed that the tracking algorithm worked well in complex environment and social occlusion case.