A method for analyzing color images containing highlight and interreflection effects

Abstract

This article studies a method for analyzing color images on non-homogeneous dielectric objects such as plastics involving highlight and interreflection effects. In the current study, the authors assumed a general case of interreflection based not only on the highlight from two adjacent surfaces of the object, but also on the specular reflection and diffused reflection. First, a model of interreflection between two planes was generated, and it was demonstrated that the observed color vector can be expressed as a linear combination of four components: the light source color, the colors of the two objects, and the diffused interreflection color. The component estimation was realized in three steps. In the first step, the light source color was estimated from the two highlight areas. In the second and third steps, respectively, chromaticities of the object colors and the diffused interreflection color were estimated. This article includes a study of the color distribution with interreflection effects using the chromaticity plane. In this study, the objects' colors were determined from analysis of the highlight and color cluster histograms. The diffused interreflection color was obtained as the common color component of the two interreflection areas. In addition, an algorithm is offered for expanding the original image into the color components of four types. To perform this, the authors used a technique based on linear programming to determine weights of the components from the RGB data. Finally, we report on tests conducted to confirm feasibility of the proposed method of analysis. © 1998 Scripta Technica. Syst Comp Jpn, 28(13): 8–18, 1997