基于高光谱数据的土壤含盐量BP神经网络模型研究

运用Hyperion数据,以黑龙江省大庆市某一实验区为例,进行了对土壤含盐量的定量研究,将BP神经网络模型(Back Propagation Network)应用到高光谱数据对研究地区土壤含盐量的反演中。通过对隐含层的传递函数、输出层的传递函数、训练算法的优化组合以及最适合隐层节点数量,得到最优的BP神经网络模型,实现了土壤含盐量的反演。对高光谱数据反演土壤含盐量采用BP神经网络具有一定的指导意义。

Research on the BP neural network model of soil salt contents by using hyperspectral data

One experimental area in Daqing city in Heilongjiang Province is took as an example to perform the quantitative soil salt contents study by using Hyper spectral data in this paper. The BP neural network model is applied to the inversion of salt contents by hyperspectral data. The optimal BP neural network model is obtained by the hidden layer transfer function, the output layer transfer function, the training algorithm optimized combination and the number of nodes which suited the hidden layer best, and the soil salt contents is inversed by using the model. The experimentent has certain guidance on the BP neural network model of soil salt contents by using Hyperspectral data.