Face and Facial Feature Tracking for Natural Human-Computer Interface

Abstract

A method for face and facial features tracking in a low-resolution web camera video stream is described. The detection results are: face bounding ellipse, eyebrow line, nostrils and mouth position. The method works at reasonable framerates on a low-quality web camera on a Pill system. The implementation of the method can be used as input module for a natural human-computer interface system to provide camera-based mouse control. The techniques used are: face tracking using color, image edge maps analysis, variant of Hough transform, template matching, color-based image segmentation.