Consumer products user interface using face and eye orientation

Abstract

In this paper, we propose an estimate of face orientation by detecting the eyes and mouth on a facial image. Our purpose is to develop a new vision-based human interface technology that uses face orientation to interact with humans. Our algorithm samples a range of skin colors and dynamically updates feature templates from the current input image. In addition, we estimate face orientation from the relationship between the detected facial features' positions. As a result, our system determined the correct location of both the eyes and the mouth for all four subjects more than 94% of the time.