Estimating Spectral Reflectance of Fabrics from High-Resolution Multi-band HDR Images(<Special issue>Digital Museum Evolution)

Abstract

In this paper, we propose a new method to estimate spectral reflectance of glossy objects such as silk-like fabrics from high resolution multi-band HDR images. Specular reflection of fabrics is known to be body-colored and quite larger than diffuse reflection, which makes spectral reflectance estimation unstable. To overcome this problem, luminance of multi-band signal is normalized before estimating spectral reflectance estimation. Experimental results on glossy color charts and a satin fabric demonstrated superior performance over conventional methods. The proposed method was also effective regardless of the estimation methods such as PLS and Wiener estimation, multiple regression analysis.