基于线性混合模型/神经网络模型的高光谱混合像元分解算法比较-以徐州地区Hyperion影像为例
首发时间:2011-02-04
摘要:混合像元的普遍存在使得传统的基于像元级的遥感分类和面积测量很难达到使用要求的精度,所以混合像元的分解问题就成为了关键。本文研究了线性混合光谱模型与BP神经网络模型,在对徐州地区的Hyperion影像进行预处理的基础上,基于MATLAB平台来进行混合像元分解实验,应用均方根误差来评价模型的性能。实验结果表明:两种混合模型均取得理想的分解结果,对数据进行进一步分析表明,对于本文数据徐州地区的Hyperion影像线性光谱混合模型的精度略高于BP神经网络模型,但是BP神经网络模型在细节信息提取方面的效果要好于线性光谱混合模型。
关键词: 混合像元分解 线性混合模型 神经网络模型; 均方根误差
For information in English, please click here
A Comparison of the Algorithm for Hyperspectral Mixed Pixel Decomposition based on Linear Mixing Spectral Model/Neural Network Model-A Case Study of Xuzhou Hyperion Image
Abstract:The mixed pixels widely existing in remote sensing made the accuracy of normal classification and area measure based pixels always difficultly to meet using needs ,so it bacame the key of the decomposition of mixed pixels problem . This paper studies the linear mixing spectral model and BP network model . Taking the preprocessed EO-1 Hyperion image of xuzhou as the experimental data, special mixture analysis is achieved based on Matlab platform, and apply the RMSE model to evaluate the two models. The results indicates that in this experiment,this two models can both achieve good decomposition results. Further analysis showed that , for the xuzhou hyperion image linear mixing model effect is better than BP neural network model. But in detail information extraction effect the BP neural network model is better.
Keywords: mixed-pixel decomposition Linear mixing spectral model neural network model root mean square error
论文图表:
引用
No.4408268468633129****
同行评议
共计0人参与
勘误表
基于线性混合模型/神经网络模型的高光谱混合像元分解算法比较-以徐州地区Hyperion影像为例
评论
全部评论0/1000