A method is introduced for acquiring the spectral image of a three-dimensional object by means of a multi-channel camera and estimating various reflection parameters of the object surface. It is assumed that the surface material is nonmetallic inhomogeneous dielectric. The Torrance-Sparrow model is used as a 3D light reflection model for describing the reflection properties. The camera system is a six-color camera, composed of a monochrome CCD camera, six color filters, and a personal computer. Next, we describe algorithms for estimating the reflection parameters from a single spectral image. The parameters estimated are both spectral functions of surface-spectral reflectance and illuminant spectral distribution, the coefficient of body-to-interface ratio, and the geometric parameter of surface roughness. Finally, experiment results are shown using a paint object. The feasibility of the method is confirmed, and the estimated model parameters are used for creating realistic computer graphics images.
View full abstract