复合神经网络在混浊介质光散射参数确定中的应用

Application of Compound Neural Networks in Optical Properties Determination from Backscattering Light of Turbid Media

为从无偏振混浊介质背向散射光确定散射光学参数,设计了倾斜入射的模拟光路以及用复合反向传播神经网络求解的方案。通过与现存的正入射-数据匹配测量方案和单一反向传播神经网络的对比,结果表明,倾斜入射-复合反向传播神经网络方案在精度和适应能力上是当前各种方案中最优的。

To determine the optical parameters of turbid media with an unpolarized backscattering light, an oblique incidence geometry with the compound BP neural network has been designed. Compared with the present scheme based on normal incidence, angular scattering measurement and data-matching and the one with single BP neural network, we conclude that this scheme is one of the most optimum scheme in the aspects of accuracy and reliability at present.