基于Fuzzy-ARTMAP网络的高光谱遥感图像分类方法

Hyperspectral imagery classification based on a Fuzzy-ARTMAP neural network

神经网络是一种重要的高光谱图像分类方法。本文提出了四边形隶属函数和Fuzzy-ARTMAP神经网络相结合的高光谱图像分类方法。该方法不局限与隶属度函数的选择,使网络更具有广泛的适用性。将其应用于高光谱图像的分类中,其分类精度高于ARTMAP神经网络,且性能稳定。

Neural network is one of methods in remote sensing image classification. This paper presents a classifica, tion method that combines quadrilateral membership function with fuzzy - ARTMAP neural network. This method isn＇ t limited to choosing the membership function. So it make the applicability of network extensive. This method is used in the hyperspectral remote sensing image classification. The accuracy of classification is higher than ARTMAP network＇ s.