基于Landsat遥感影像的杨树信息提取方法分析

Study on information extraction of poplar resources using Landsat remotely sensed datasets

该文使用近10 a 4个时相的江苏全省Landsat遥感影像,在数据预处理的基础上提取归一化植被指数、比值植被指数、土壤调节植被指数、增强型植被指数、大气阻抗植被指数等5种植被指数,并进行主成分分析。运用最大似然法、随机森林法和光谱角填图法进行分类,结合小班数据,对各方法的分类结果进行精度评价。评价结果表明,光谱角填图分类法在杨树信息提取时精度更高,对杨树的区分精度也达到42.67%。

In this article,five kinds of vegetation index, the first principal component and five kinds of texture features （NDVI,SR,SAVI,EVI,ARVI） were extracted on the basis of data collection from and pretreatment of Landsat remote sensing images during ten years （from 2005 to 2015） in Jiangsu Province. In combination with field survey data and kappa coefficient index analysis,three classification methods as the maximum likelihood method,the random forest method and spectral angle mapping were compared for accuracy assessment. Result showed that the spectral angle mapping method got the highest accuracy in poplar information extraction,with accuracy of 42. 67%.