线性波谱分离在城市土地覆盖影像研究中的应用  

Application of linear spectral unmixing in the study of urban land cover images

作　　者：陈有明[1] 刘琳[2] 肖正辉[3] 李学良[3] 刘文长[4] 刘瑱 

机构地区：[1]安徽省地质调查院,安徽合肥230001 [2]安徽农业大学理学院,安徽合肥230036 [3]合肥工业大学化学工程学院,安徽合肥230009 [4]安徽省地质实验研究所,安徽合肥230001

出　　处：《合肥工业大学学报（自然科学版）》2013年第3期365-368,共4页Journal of Hefei University of Technology：Natural Science

基　　金：安徽省国土资源厅科技资助项目(2010-g-19;2011-K-11)

摘　　要：文章以合肥市区TM影像为数据源,采用非负最小二乘算法,解决ENVI等商业软件在该算法上存在的不足;对数据进行线性波谱分离,通过对丰度总和数据和残余均方根误差等指标的分析,对TM影像进行了解析;对于TM图像土地覆盖中的混合像元限制分类精度的问题,采用对TM图像MNF变换结果的前5个主成分进行PPI指数计算,并对PPI阈值切割结果进行N维可视化;采用波谱均值作为端元波谱,选择端元数为5,对2种端元类型方案、4种不同的波谱特征进行数值处理,并以模型自身的条件约束等进行验证和样地数据的检核,发现该方法的有效性。The non-negative least-square algorithm was employed to accomplish linear spectral unmixing（LSU） with Hefei TM images as data source to overcome the inadequacy of the ENVI software.The TM images were resolved based on the analysis of the data abundance,sum and residual root mean square error.In view of the accuracy problem of mixed pixel limit classification in TM images for land cover,PPI index of the first five principal components of the results of TM image MNF transform was calculated and N dimension visualization was performed on PPI threshold cutting results.The numerical process was performed on two types of end member scheme and four kinds of spectral characteristics using spectral mean as end member spectrum and selecting five end members.It is found that the method is effective for the results were confirmed by the constraint condition and checked by sample data.

关 键 词：线性波谱分离 非负最小二乘算法 合肥 混合像元 土地覆盖 端元 

分 类 号：S159[农业科学—土壤学] TP79[农业科学—农业基础科学]