Lu Xiang, Gross Herbert
Opt Express. 2020 Dec 21;28(26):39368-39386. doi: 10.1364/OE.411028.
Simulation of scattering from optical surfaces is usually based on Monte Carlo methods in which the bidirectional scattering distribution function (BSDF) of the optical surfaces are sampled randomly by many rays, resulting in long calculation times. In order to accelerate the simulation, a quasi-analytical phase space model is proposed. In this model, few rays are traced from the object and image space to the target surface to determine the illumination and acceptance areas in phase space, where these areas can be conveniently coupled simultaneously in the spatial and angular domain. Since no random sampling is involved in the phase space model, no statistical noise perturbs the result and the surface scattering simulation can be greatly accelerated. Additionally, due to the use of real raytracing, the phase space model removes the limitation of paraxial approximation, which usually limits the accuracy of deterministic stray light analysis models. Meanwhile, by the discretization of the optical surfaces into subareas, this new approach is able to model freeform surfaces with arbitrary geometries, and space-variant BRDFs can be applied for different subareas of the optical surface.
光学表面散射的模拟通常基于蒙特卡罗方法,其中通过许多光线对光学表面的双向散射分布函数(BSDF)进行随机采样,这导致计算时间较长。为了加速模拟,提出了一种准解析相空间模型。在该模型中,从物空间和像空间向目标表面追踪少量光线,以确定相空间中的照明和接受区域,这些区域可以在空间和角度域中方便地同时耦合。由于相空间模型不涉及随机采样,因此没有统计噪声干扰结果,并且可以大大加速表面散射模拟。此外,由于使用了实际光线追踪,相空间模型消除了傍轴近似的限制,傍轴近似通常会限制确定性杂散光分析模型的精度。同时,通过将光学表面离散为子区域,这种新方法能够对具有任意几何形状的自由曲面进行建模,并且空间变化的BRDF可以应用于光学表面的不同子区域。