基于Landsat?8遥感影像的湿地覆被监督分类研究

Land?cover Supervised Classification Research Based on Landsat?8Remote Sensing Data

基于监督分类法在遥感影像分类中的普遍应用,首先将研究区分为水体、养殖塘、碱蓬、芦苇、水田、滩地、居民点、其他等8个类型,然后采用辽宁省双台子河口湿地2014-05-26的Landsat-8遥感影像数据,运用ENVI提供的常用的监督分类方法——最大似然法对研究区的遥感影像进行湿地覆被分类研究,并将研究区2014-05-18的ETM+影像的分类结果与其进行对比分析。最后,利用混淆矩阵分别对不同影像的分类结果进行精度评价,比较了两者的准确程度。结果表明:Landsat-8OLI影像的分类总体精度达到了92.25%,比Landsat-7ETM+影像分类总体精度高11.46%,能够更好的进行湿地覆被信息提取,同时验证了基于Landsat-8遥感影像湿地覆被分类的可行性,扩充了Landsat-8遥感影像的应用范围。

Supervised classification is widely applied in remote sensing classification research .Firstly ,the study area was divided into eight categories ,including upland ,water body ,culture pond ,suaeda ,reed ,,upland ,beach ,residential land and remainder .Second‐ly ,land-cover information in Shuangtaizi estuary wetland was extracted with the classifier of maximum likelihood provided by EVNI 5 .0 software using Landsat-8 remote sensing data on May 26 ,2014 and ETM＋ remote sensing data on May 18 ,2014 .Finally ,ac‐curacy assessment was conducted with confusion matrix .Then this paper compared the accuracy between the two images .Results show that classification overall accuracy of Landsat-8 OLI is 92 .25% ,which is 11 .46% higher than ETM＋ .It is proved that Land‐sat-8 is feasible in wetland land-cover information and the scope of application is wider .