基于实测光谱对喀斯特地区土地覆盖类型丰度提取  

Extract the abundance of land cover types based on the measured spectra of Karst region

作　　者：李岩[1,2] 安裕伦[1,2] 李晓宇[3] 吴松[1,2] 

机构地区：[1]贵州师范大学地理与环境科学学院,贵州贵阳550001 [2]贵州师范大学贵州省山地资源与环境遥感应用重点实验室,贵州贵阳550001 [3]中国人民解放军总参谋部第一测绘导航基地遥感测绘院,辽宁大连116000

出　　处：《贵州师范大学学报（自然科学版）》2015年第4期9-13,共5页Journal of Guizhou Normal University：Natural Sciences

基　　金：贵州省省长专项基金项目(黔省专合字(2011)46号);贵州省科技厅攻关项目(黔科合GY字(2007)3017);国家自然科学基金项目(41161002)

摘　　要：利用线性光谱模型对土地覆盖类型相对复杂的喀斯特地区进行混合像元分解。以贵州省镇宁县为研究区域,基于实测光谱信息,以Landsat TM8影像为信息源,选取裸土、裸岩、绿色植被、干枯植被为端元,利用线性分解模型提取地类覆盖信息,计算端元覆盖面积。通过对丰度图像的归一化植被指数(NDVI)值与TM8经过缨帽变换的NDVI值进行回归分析,验证端元选取精度,得到R2>0.85。研究表明:在喀斯特山区,利用实测光谱混合像元分解,有效提取植被、裸岩、裸土丰度,提高土地覆盖分类精度的方法是可行的。The study decompose mixed pixels of karst region that the land cover types relatively complex by using the linear spectral model,and furhter extract on the proportion of vegetation,bare soil,and bare rock. This paper chose the Zhenning county of Guizhou province as the case study,based on the measured spectral information,selected the bare soil,bare rock,green vegetation and dry vegetation as the endmember and used the linear spectral model to extract land cover information,calculated the coverage areas. The result is R^2〉 0. 85 through regression analysis of NDVI of abundance image and TM8 of tasseled cap transfermation. The results show that using the measured spectral decomposition of mixed pixels can effectively extract the abundance of vegetation,bare rock and bare soil and to improve the classification accuracy of land cover.

关 键 词：实测光谱 混合像元 喀斯特 线性光谱混合模型 

分 类 号：P237[天文地球—摄影测量与遥感]