On Reducing the Complexity of Illumination Cones for Face Recognition

In our previous work [1], we have demonstrated how to arrange physical lighting so that the acquired images of a human face can be directly used as the basis vectors of a lowdimensional linear space. The “universal configuration” computed in [1] has been shown to provide a good basis for face recognition under a wide range of lighting conditions. In this paper, we propose a more rigorous formulation of the problem of computing the universal configuration. We propose a new objective function Ob and define the domain of Ob, in which we seek our universal configuration. Recognition experiments and other results reported in the paper have validated this new approach.