Face recognition on Lorentzian manifold

Face recognition is one of the hottest topics of pattern recognition. Many linear and nonlinear models are developed for face recognition. Beside different studies in literature, linear methods with manifold learning has an important potential. In this work, a new approach on Lorentzian geometry is presented. After classes and data are modeled on Lorentzian manifold, data sets on new manifold are characterized using Lorentzian metric tensor in this approach. Face classification is done accurately on new space which is transformed from sample space by Lorentzian metric tensor.