Stochastic modeling of non-Lambertian surfaces for Monte Carlo computations in optical radiometry

摘要

The accuracy of Monte Carlo calculations in optical radiometry is often limited by the reflection model employed (usually, diffuse (Lambertian), or simplified specular-diffuse). There are many cases when these models cannot adequately simulate the bidirectional reflection distribution functions (BRDFs) of real materials, which can call into question the results of the calculations. The features and the methods of modeling with the use of importance sampling for various materials are considered with a focus on an acceptance-rejection technique and on a procedural ab initio method. The Monte Carlo algorithms for modeling non-Lambertian reflection are discussed. The simulated and measured BRDFs for several materials are compared.