基于最优波段组合的土地利用/覆盖遥感信息提取研究  

Study on the Extraction of Remote Sensing Information of Land Use/Cover Based on Optimum Band Combination

作　　者：李谢辉[1,2] 郑奕[3] 

机构地区：[1]河南大学黄河文明与可持续发展研究中心,河南开封475001 [2]兰州大学西部环境教育部重点实验室,甘肃兰州730000 [3]中国气象局乌鲁木齐沙漠气象研究所,新疆乌鲁木齐830002

出　　处：《安徽农业科学》2009年第14期6696-6699,共4页Journal of Anhui Agricultural Sciences

基　　金：教育部人文社会科学重点研究基地重点项目(YRCSD08A05);国家自然科学基金项目(40501077)

摘　　要：以渭河下游河流沿线区域为研究区,通过对2002年ETM+影像的各波段光谱特征、相关系数矩阵、最佳指数OIF、修正植被指数RNDVI、主成分变换进行分析后,认为第一主成分分量PC1、RNDVI和Band4为最优波段组合。利用非监督的ISODATA和监督分类相结合的方法对研究区进行土地利用/覆盖分类后,得到的总体分类精度为90.098 7%,Kappa系数为0.884 5,说明该研究方法获得的分类精度较高,步骤简便且实用性强,能极大地提取遥感分类信息。Taking the catchment area along Weihe River in Shaanxi Province as a study area, by analyzing spectral characteristics of every band, correlation coefficient matrix among bands, optimum index, revised vegetation index and principal components transformation on ETM ＋ image in 2002, optimum band combinations were fixed on PCl, RNDVI and Band4. After classified to study area on land use and cover using unsupervised and supervised classification, the total accuracy of classification was 90. 098 7% , and the Kappa coefficient was 0.884 5. The results show that the method not only gets higher accuracy, but has simpler steps and stronger practicability, and can extract furthest remotelysensed classified information.

关 键 词：最佳指数 修正植被指数 主成分分析 非监督分类 监督分类 

分 类 号：F301.2[经济管理—产业经济]