Face recognition with multimodal subspace analysis

Appearance based methods happen to be very useful for general face recognition tasks. The main problem with this methods is the multimodality of images under changing variations. In this study, we propose a novel approach which can be utilized against the multimodality. The technique, that is based on the nonlinear manifold embedding, defines a linear subspace for illumination variations. We obtained some promising recognition results which are comparable with state of art methods.