Recovering 3D Shape and Albedo from a Face Image under Arbitrary Lighting and Pose by Using a 3D Illumination-Based AAM Model

摘要

We present a novel iterative approach for recovering 3D shape and albedo from face images affected by non-uniform lighting and non-frontal pose. We fit a 3D active appearance model based on illumination, to a novel face image. In contrast to other works where an initial pose is required, we only need a simple initialization in translation and scale. Our optimization method improves the Jacobian each iteration by using the parameters of lighting estimated in previous iterations. Our fitting algorithm obtains a compact set of parameters of albedo, 3D shape, 3D pose and illumination which describe the appearance of the input image. We show that our method is able to accurately estimate the parameters of 3D shape and albedo, which are strongly related to identity. Experimental results show that our proposed approach can be easily extended to face recognition under non-uniform illumination and pose variations.