Effective compression and reconstruction of human skin hyperspectral reflectance databases

In this paper, we investigate techniques for reducing the dimensionality of skin hyperspectral reflectance databases and maintaining a high degree of fidelity during data reconstruction. We compare results obtained using principal components analysis (PCA) with results provided by a piecewise PCA approach that explores the different roles performed by the main light attenuation agents acting within the cutaneous tissues in the ultraviolet (UV), visible and near-infrared (NIR) domains. Our investigation encapsulates not only skin spectral responses obtained by varying the contents of these agents, but also responses resulting from the absence of melanin pigmentation associated with the vitiligo condition.