基于神经网络的LAI遥感反演影响因素分析

Analysis of Influence Factors for LAI Inversion Based on Neural Network

基于BP神经网络和植被冠层辐射传输模型PROSAIL对LAI进行反演,针对BP神经网络中网络结构设计的关键问题,从网络隐含层层数、隐含层节点数和学习速率等3个方面进行分析与讨论,确定了最佳的BP网络结构,取得了较好的反演结果。

In this paper,we discussed a LAI inversion method based on BP neural network and vegetation canopy radiative transfer model PROSAIL at first.And then,we analyzed the key issues of BP neural network structure design,from the number of hidden layers,the number of hidden layer nodes and learning rate.In the end,we determined the optimal structure of BP network,and obtained a better result.