基于混合粒子群算法的目标表面BRDF建模与优化

Target Surface BRDF Modeling and Optimization Based on Hybrid Particle Swarm Optimization

针对目前流行的目标表面双向反射分布函数(BRDF)模型建立多基于大量实验数据且存在优化算法精度不高的问题,提出基于混合粒子群算法的目标表面BRDF建模与优化方法。该方法将模拟退火引入粒子群算法,构建了模拟退火-粒子群算法(SA-PSO)用于求解四参数单站(FOI)BRDF模型,四参数即镜面反射幅度系数、镜面反射系数、漫反射幅度系数和漫反射系数,获得目标板BRDF模型。验证计算结果表明:SA-PSO计算能力优于拟牛顿算法和改进信赖域算法且适用于多种BRDF模型求解,建立的BRDF模型与真实值最小均方差为0.72%,可为目标表面的反射特性研究提供参考。

The current popular target surface bidirectional reflectance distribution function(BRDF)model is based on a large number of experimental data and the accuracy of the optimization algorithm is not high.In view of the above problems,the relative measurement method was used to obtain the BRDF value of the target plate.The characteristics of the target surface BRDF model were analyzed.Referring to the four-parameter single station(FOI)BRDF model,a simulated annealing-particle swarm optimization(SA-PSO)algorithm was proposed to optimize the four parameters of the FOI model,namely the mirror reflection amplitude coefficient,the mirror reflection coefficient,the diffuse reflection amplitude coefficient and the diffuse reflection coefficient.It was indicated that the calculation ability of SA-PSO was better than that of quasi-Newton algorithm and improved trust region method,and it was suitable for solving various BRDF models.The minimum mean square deviation between the BRDF model established by this method and the real value was 0.72%.Taking into account the measurement error and other factors,the accuracy was high,which could provide reference for the study of the reflection characteristics of the target surface.