COLOR REFLECTANCE MODELS FOR RECOGNTION AND COMPUTER GRAPHICS
Glenn E. Healey · 2005
Abistract We describe a system for representing the physical properties of maaterial surfaces using general physical models. We accurately model the optical properties of both homogeneous and inhomogeneous materials. Our model for the scattering of light from the body of inhomogeneous materials based on modified Kubelka-Munl; theory is inore general than previous models used in computer graphics for this process. It is this scattering that is largely responsible for the appearance of most of the objects we see in everyday life. We unify our physical and geometric representations using a method to map material surface properties onto our geometric models. Our system contains a flexible geometric modeling component which is used to represent objects. We represent the physical properties of material surfaces using a generic physical modeling system. Every parameter of our representation specifies an intrinsic physical property of a material surface. We represent a material surface ils the combination of a inaterial description and a surface description. Physically, the material description specifies optical properties while the surface description specifies local geometric properties. Our models apply to both lioniogeneous and inlio~nogeneous materials. While most of the work om reflectance models in coinputer graphics has concentrated on specular reflection, we adopt here a new physical model for a process which is much inore significant visually. This process is called colorant layer scattering and it is the dominant optical process for inhomogeneous materials (e.g. plastics, paints, textiles, paper, ceramics). The model we use has been verified experimentally by others and allows us to accurately predict variations in tlie color and intensity of the light scattered from the body of an inlionioge~~eous material as a function of geometry. Our system also includes general models for liglit sources. The pliysical properties (intensity and color) of the light reflected from surfaces are easy to compute and render using our physical models. We merge our geometric and physical representations for objects using functions which associate material and surface properties witli regions on geometric models.