Pose Invariant Face Recognition Under Arbitrary Unknown Lighting Using Spherical Harmonics

Abstract. We propose a new method for face recognition under arbitrary pose and illumination conditions, which requires only one training image per subject. Furthermore, no limitation on the pose and illumination conditions for the training image is necessary. Our method combines the strengths of Morphable models to capture the variability of 3D face shape and a spherical harmonic representation for the illumination. Morphable models are successful in 3D face reconstructions from one single image. Recent research demonstrates that the set of images of a convex Lambertian object obtained under a wide variety of lighting conditions can be approximated accurately by a low-dimensional linear subspace using spherical harmonics representation. In this paper, we show that we can recover the 3D faces with texture information from one single training image under arbitrary illumination conditions and perform robust pose and illumination invariant face recognition by using the recovered 3D faces....

• Type: Conference
• Year: 2004
• Where: ECCV
• Authors: Lei Zhang 0002, Dimitris Samaras
• Keywords: Arbitrary Illumination, Computer Vision, ECCV 2004, Illumination Conditions, Images