Reflectance spectra of 3160 specimens from the Munsell and the NCS atlases have been used as a model group in PCA and we have verified that a basis of six eigenvectors, without mean vector, is sufficient for a good spectra reconstruction. The use of the PCA provides good reconstructions but has the drawback of not allowing the reconstruction of spectra of specimens from other sample sets different from the model group. If the results are to be applied to spectra of samples other than those used previously, it is necessary to calculate the corresponding principal components in a different way. The use of six tristimulus values from each spectrum measured, obtained using the CIE D65 and A illuminants, in the calculation of the principal components, Mi, permits the reconstruction of any spectrum, while providing, in addition, a high degree of metamerism between the measured and the reconstructed spectra.
Angel I. Negueruela, José F. Echávarri, "Reconstruction of surface reflectance spectra by means of tristimulus values" in Proc. IS&T CGIV 2008/MCS'08 4th European Conf. on Colour in Graphics, Imaging, and Vision 10th Int'l Symp. on Multispectral Colour Science, 2008, pp 104 - 107, https://doi.org/10.2352/CGIV.2008.4.1.art00024