NWFE结合纹理特征的SVM土地覆被分类方法研究  

Land Cover Classification with SVM Based on NWFE and Texture Features

作　　者：崔林林[1,2] 罗毅[1] 包安明[1] 

机构地区：[1]中国科学院新疆生态与地理研究所,乌鲁木齐830011 [2]中国科学院研究生院,北京100049

出　　处：《国土资源遥感》2012年第1期36-42,共7页Remote Sensing for Land & Resources

摘　　要：为提高土地覆被分类精度,采用非参数权重特征提取(nonparametric weighted feature extraction,NWFE)结合纹理特征的支持向量机(support vector machines,SVM)的分类法,对新疆玛纳斯河流域绿洲区2006年的土地覆被进行分类,并将该方法与主成分分析(principal component analysis,PCA)结合纹理特征的SVM分类、原始波段结合纹理特征的SVM分类进行对比。结果表明,NWFE结合纹理特征的SVM分类结果优于其他2种分类结果,不仅反映了土地覆被分布的整体情况,而且使不同土地覆被类型得到较好的区分,总体分类精度达89.17%。Land cover classification based on remote sensing image is of significant importance to agriculture,forestry and environment monitoring.Algorithm of remote sensing information retrieval is always an important research topic in this field.This paper made an effort to combine the Nonparametric Weighted Feature Extraction（NWFE） and texture features with the Support Vector Machines（SVM） so as to achieve a higher classification precision.The combined approach was applied to land cover classification of the Manasi River oasis in Xinjiang in 2006,and was compared with approaches of SVM based on Principal Component Analysis（PCA） and texture features and based on original bands and texture features.The results show that the method of SVM combined with NWFE and texture features can capture not only the distribution of land cover but also the difference among land cover types.An overall classification accuracy of 89.17% is obtained,which is better than those of two other classification results.

关 键 词：非参数权重特征提取(NWFE) 支持向量机(SVM) 土地覆被分类 

分 类 号：TP751.1[自动化与计算机技术—检测技术与自动化装置] TP274[自动化与计算机技术—控制科学与工程]