Head Pose Computation for Very Low Bit-rate Video Coding

In this paper we examine the problem of determining 3D human head pose from a sequence of 2D grey scale images. Computation of head pose is vital in many areas of human head modeling and applications include model-based video compression, face recognition, 3D head tracking, and Human Computer Intelligent Interaction (HCII). A significant amount of research has been done in the general area of 3D object alignment for recognition and modeling purposes with much of the work concentrating on applying techniques to simple polyhedral objects such as cubes or polyhedrons. We take these techniques a step further and apply them to grey scale images of human heads using a set of distinguishing facial features. Using at least 3 of these features visible in each 2D image, we compute the pose relative to a given 3D model. We optimize the selection of feature points by minimizing the error in mapping the remaining points with the computed pose. The results we obtain for a wide range of head orientations and scales proved to be quite accurate. In addition, the computational efficiency of the system is high enough to be used in real-time head tracking or a model-based video coding system.