Head Pose Estimation Based on Manifold Embedding and Distance Metric Learning

Xiangyang Liu, Hongtao Lu, Daqiang Zhang

Abstract:

In this paper, we propose an embedding method to seek an
optimal low-dimensional manifold describing the intrinsical pose varia
tions and to provide an identity-independent head pose estimator. In
order to handle the appearance variations caused by identity, we use a
learned Mahalanobis distance to seek optimal subjects with similar man
ifold to construct the embedding. Then, we propose a new smooth and
discriminative embedding method supervised by both pose and identity
information. To estimate pose of a head new image, we first find its k
nearest neighbors of different subjects, and then embed it into the man
ifold of the subjects to estimate the pose angle. The empirical study on
the standard databases demonstrates that the proposed method achieves
high pose estimation accuracy.