Human tracking with multiple cameras based on face detection and mean shift

Human tracking is an important function to an automatic surveillance system using a vision sensor. Human face is one of the most significant features to detect person(s) in an image. However, face is not always observed from a single camera. Therefore, it is difficult to identify a person exactly in an image due to the variety of poses. This paper describes a method for automatic human tracking based on the face detection using Haar-like features and the mean shift tracking method. Additionally, the method increases its trackability by using multi-viewpoint images. The validity of the proposed method is shown through experiment.