Efficient real-time face tracking in wavelet subspace

Abstract

We present a new method for visual face tracking that is carried out in wavelet subspace. First, a wavelet representation for the face template is created, which spans a low-dimensional subspace of the image space. The video sequence frames where the face is tracked are then orthogonally projected into this low-dimensional subspace. This can be done efficiently through a small number of applications of the wavelet filters. All further computations are performed in wavelet subspace, which is isomorphic to the image subspace spanned by the sets of wavelets in the representation. Robustness with respect to facial expression and affine deformations, as well as the efficiency of our method, are demonstrated in various experiments.