Illumination Compensation and Enhancement for Face Recognition

Face images of the same person under different illuminations represent a challenge for face recognition. Traditional methods based on the Lambertian model construct a face image invariant to illuminations by combining a number of the images of different illuminations linearly. A drawback with the Lambertian model is that a single-point light source placed at infinity is assumed. This paper proposes an efficient scheme for illumination compensation and enhancement of face images. Our illumination model is universal without requiring the assumption of a single-point light source, so it circumvents and overcomes the limitations of the Lambertian model. In practice, it is reasonable to assume that the variations in intensity and illumination directions cause the face images of the same person different. The proposed approach can learn the average representations of face images under changing illuminations so as to compensate or enhance the face images and to eliminate the effect of different and uneven illuminations while keeping the intrinsic properties of the face surface. Our experiments have provided promising results, demonstrating that the proposed methods are effective.