Recovering Material Reflectance from Polarization and Simulated annealing

Abstract : This paper presents a novel image-based method for estimating BRDF information. The goal is to determine the dependence of the reflected radiance on the surface orientation for any given illumination direction. The proposed method consists of two stages. In the first stage, the polarization state of light reflected from an object of unknown shape is analysed using a linear polarizer and digital camera. This is interpreted using Fresnel theory to estimate a field of surface normals. This stage of the algorithm is largely based on existing ideas. In the second stage, a 3D histogram of surface normals and measured pixel brightnesses is constructed. The BRDF is recovered by fitting an arbitrary surface to this histogram data using simulated annealing and elastica. The experiments presented in this paper show that realistic image rendering is possible based on the recovered BRDFs of smooth dielectric materials.