Numerical reconstruction of spectral reflectance curves of oil painting on canvas

摘要

Unlike the color -which is a quality without physical meaning and that involves subjective estimation of interaction of electromagnetic radiation with surfaces- spectral reflectance is a physical property that characterizes different materials, no matter what chromatic content of illuminant and the spectral response of the sensor. This means that the spectral reflectance is a magnitude of particular interes in both reconstruction and reproduction in digital color systems. In this paper, two approaches to the numerical reconstruction of spectral reflectance curves of samples of oil painting on canvas, are presented. These approaches need a set of spectral reflectance curves, given by a spectrophotometer, and their respective sampling using color filters placed in front of a monochrome CCD camera. The first approach is based on the interpolation of the camera response to each color filter. The second one, relies in obtain a vectorial base and appropiate coefficients to reconstruct the spectral reflectance curve. Goodness of fit coefficient (GFC) and absolute mean error (ABE) are the metrics used to evaluate the performance of the proposed procedures.