Landsat-8遥感影像不同水体提取方法适用性分析

Applicability analysis of different water extraction methods from Landsat-8 remote sensing images

为探究周围气温与地物环境对水体提取的影响,利用Landsat-8 OIL遥感影像,选取巢湖作为研究区,分别采用水体提取指数模型(NDWI)、监督分类最大似然法及面向对象分类法对不同季节下的研究区做水体提取。通过分析比较各方法在不同季节对水体提取的精度影响,发现面向对象法提取效果最佳,NDWI效果次之,监督分类精度最低。监督分类和归一化水体指数对四季水体提取精度较平均,冬季用面向对象法提取水体影响大。

In order to explore the influence of ambient temperature and surface environment on water extraction,Chaohu Lake is selected as the research area by using landsat8 Oil remote sensing image.Water extraction index model(NDWI),supervised classification maximum likelihood method and object-oriented method are used to extract water in different seasons.By analyzing and comparing the influence of each method on the accuracy of water extraction in different seasons,it is found that the object-oriented method has the best extraction effect,the NDWI effect is the second,and the supervised classification accuracy is the lowest.Supervised classification and normalized water body index have an average extraction accuracy in four seasons,and winter has a great impact on water extraction by object-oriented method.The experiment provides a reference of selecting different methods for water extraction from landsat-8 remote sensing image in different seasons.