ABSTRACT
In this chapter, we combine the intensity model of Chapter 3 and the warping model of Chapter 4 to build a model for faces that allows recognition under a wide range of lighting conditions. Our purpose here is to show that each component of the face model-shading, global warping and local warping-is necessary for recognition, and that the face model as a whole is sufficient for successfully recognizing faces under a wide range of lighting conditions. We do not claim that the algorithm used to minimize the fit of the deformable template is optimal by any criteria, but only that it is sufficient given suitably good initial conditions.
