Automatic skin-color distribution extraction for face detection and tracking

For face detection and tracking in a video sequence, many systems are now applying a technique to extract skin-color areas of faces. In these systems, the skin-color model should be built beforehand. However the skin-color changes when the lighting condition varies, when an input video camera is changed, and when the white-balance of the camera is readjusted, etc. We describe a method to a build a skin-color model for operation while a face detection system is running. When it has no skin-color model, it uses the differences between adjacent frames to extract moving objects. Once the system recognizes that the moving object is a face, it analyses the color histograms of the moving area and extract skin-color distribution. We applied this method to our face detection and tracking system. Experiments showed good results