Inverse Reflectometry for Real Objects with Diffuse and Specular Interreflections

In the study of composite reality, in which virtual objects are synthesized and displayed in a real environment, intensive research is under way on virtual reality in order to achieve the virtual representation of real objects with a greater resemblance to reality. The estimation of the surface reflection characteristics of the object is especially important in order to eliminate inconsistencies of shadow between the real environment and the virtual object. To deal with this problem, we are continuing the study of dense estimation of the reflection characteristics of objects from distant images and surface texture images. For objects of complex shape, we have proposed a method of taking account of the effect of mutual reflection due to the secondary reflections on the object surface. However, there remains the problem that the effect of mutual reflection by specular-reflection components cannot be eliminated. Consequently, a method is proposed in this paper which estimates the reflectance with allowance for the effect of mutual reflection, including the specular reflection component on the object surface. Specifically, the method is as follows. First, the illumination conditions with regard to the direction of illumination are determined from the object shape in order to observe the reflection component at a large number of points on the object surface. Based on the determined multiple illumination conditions and the measured data, photon mapping is applied, using the photon distribution on the object surface to estimate the surface reflectance at each point on the surface. In experiments, the estimated reflectance is compared to the results obtained by conventional methods, and it is shown that the proposed method can represent objects faithfully under various illumination conditions in the virtual environment. © 2006 Wiley Periodicals, Inc. Electron Comm Jpn Pt 2, 90(1): 50–60, 2007; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecjb.20324