摘要
【目的】针对传统遥感图像分类方法精度低的缺点,运用基本竞争型神经网络模型对TM影像进行分类研究。【方法】在考虑TM影像光谱信息和地表结构变化信息的基础上,应用经过基本竞争型神经网络训练后的分类器对TM影像进行分类研究,并与利用最大似然法的分类结果进行比较。【结果】研究区TM影像采用基本竞争型神经网络进行分类的总体分类精度和Kappa系数分别为89.1%和0.873,而采用最大似然法分别为70.6%和0.646,前者的分类精度明显高于后者。【结论】基本竞争型神经网络的分类结果明显优于最大似然法的分类结果。
【Objective】 Given the shortage of low classification precision in traditional remote sensing classification,the Basic Competitive NN is applied to classify TM remote sensing image.【Method】 Considering the spectral information of TM image and the changing information of surface structure,TM image classification is carried out using the classifier trained by the Basic Competitive NN,and the result from Basic Competitive NN is compared with it from MLC.【Result】 The total classification precision and Kappa index of TM image with Basic Competitive NN in the study area are 89.1% and 0. 873, while these with MLC are 70.6% and 0. 646. [Conclusion] The result shows that the classification result of Basic Competitive NN is better than that of MLC.
出处
《西北农林科技大学学报（自然科学版）》
CSCD
北大核心
2009年第8期154-160,170,共8页
Journal of Northwest A&F University(Natural Science Edition)
基金
国家"863"高新技术研究与发展计划项目(2008AA10Z223)
国家自然科学基金项目(40671145)
关键词
TM影像分类
地表结构信息
基本竞争型神经网络
最大似然法
TM image classification
surface structure information
Basic Competitive Neural Network（Basic Competitive NN）
Maximum Likelihood Classifier（MLC）
