摘要
提高遥感图像分类精度一直是受到普遍关注的焦点问题。近年来 ,人工神经网络技术和分层处理技术由于它们的许多优点受到广泛欢迎。本文把这两种技术结合起来 ,提出了分层神经网络的概念 ,并基于此设计了一种分层神经网络分类算法。通过与最大似然法的对比实验表明 ,这种分层神经网络分类算法可以明显地提高分类精度 ,并对不规则分布的复杂数据具有很强的处理能力。
Improving classification accuracy is always the focus for remote sensing image classification. Now the artificial neural network technology and hierachical processing technology are widely used in remote sensing classifications because of their excellences. In this article, these two technologies are combined and a new concept of hierachical neural network is brought forward. Based on this concept, an algorithm for remote sensing image classification is developed. The experiment comparing with the MLC algorithm indicates that the classification algorithm of hierachical neural network can easily process the difficult data that distributing abnormally and improve classification accuracy distinctly.
出处
《测绘学报》
EI
CSCD
北大核心
2000年第3期229-234,共6页
Acta Geodaetica et Cartographica Sinica
基金
国家 8 63攻关!"高光谱遥感数据处理分析系统研究"项目
关键词
分层处理
神经网络
遥感图像分类
分类精度
hierachical processing
neural network
remote sensing image classification
classification accuracy