Multi-view face recognition by nonlinear tensor decomposition

We discuss a new multi-view face recognition method that extends a recently proposed nonlinear tensor de-composition technique. We use this technique to pro-vide a generative face model that can deal with both the linearity and nonlinearity in multi-view face im-ages. Particularly, we study the effectiveness of three kinds of view manifold for multi-view face representa-tion, i.e., the concept-driven, data-driven and hybrid data-concept-driven view manifolds. An EM-like al-gorithm is developed to estimate the identity and view factors iteratively. The new face generative model can successfully recognize face images captured under un-seen views, and the experimental results provide the new method is superior to the traditional TensorFace-based algorithm and the view-based PCA method. 1