Precise Estimation of Painting Surfaces for Digital Archiving

摘要

This paper proposes a method for precisely estimating the surface properties of oil paintings for digital archiving. The surface properties include surface height information, surface spectral reflectance, and reflection model parameters. We use mainly a multiband imaging system and a high-resolution RGB camera, and use locally a spectrometer and a laser-scanning meter for precise surface measurement. First, we combine the estimated surface-spectral reflectances in high resolution from the multiband system and the measurements in low resolution from the spectrometer. Second, we combine the estimated surface height in high resolution from the RGB camera and the measurements in low resolution from a laser-scanning meter. Third, we develop a region dependent rendering algorithm where appropriate reflection parameters are determined for classified regions. All estimates of the surface properties are combined for rendering realistic color images of oil paintings. The feasibility of the proposed methods is shown in experiments using real oil paintings.