Basis Functions of the Color Signal of Skin under Different Illuminants

The skin color signal is the product of skin reflectances and spectral power distribution of illuminants. In this paper we use principal component analysis to obtain basis functions for an ensemble of skin color signals and test the reconstruction goodness with different number of basis functions and types of normalization. We demonstrate the usefulness of the basis functions in two applications: (1) the reconstruction of new skin color signals which are due to the combination of two light sources that simultaneously impinge on the skin, and (2) the computation of the “skin locus” of a color camera, which is the range of possible skin chromaticities that the camera can measure.