An analytic solution for the pose determination of human faces from a monocular image

Pose determination of human faces plays an important role in face recognition, gaze tracking, human-computer interaction and virtual reality applications. This paper presents a fast and accurate vision-based approach to determine the pose of a face from a monocular image. In this approach, only the far corners of the eyes and mouth in a facial image are used. The proposed method can yield an exact and analytic solution without the requirement of using any specific 3D facial model, special marks, extra equipment or assumptions. Finally, sensitivity analysis and experimental results are provided to verify the theory. (C) 1998 Elsevier Science B.V. All rights reserved.