Fitting a Morphable Model to 3D Scans of Faces

Abstract

This paper presents a top-down approach to 3D data analysis by fitting a \nMorphable Model to scans of faces. In a unified framework, the algorithm \noptimizes shape, texture, pose and illumination simultaneously. The algorithm \ncan be used as a core component in face recognition from scans. In an \nanalysis-by-synthesis approach, raw scans are transformed into a PCA-based \nrepresentation that is robust with respect to changes in pose and illumination. \nIllumination conditions are estimated in an explicit simulation that involves \nspecular and diffuse components. The algorithm inverts the effect of shading in \norder to obtain the diffuse reflectance in each point of the facial surface. \nOur results include illumination correction, surface completion and face \nrecognition on the FRGC database of scans.