摘要
文章以"珠海一号"卫星星座02组A星(OHS-2A)高光谱影像为数据源,以云贵高原的洱海为主要研究区域,设计了一种新的水体提取指数——阴影建筑指数(SBI),该指数利用归一化差异水体指数和水体在近红外波段强吸收的特性,构建了二者的乘积关系式,拉大水体与其它地物(尤其是建筑物)的灰度差异,进而更准确地提取水体。为了显示其优越性,与单波段阈值法、多波段谱间关系法、归一化水体指数法、改进的阴影水体指数法、和综合权重水体指数法等五种已有的水体提取方法进行比较。通过混淆矩阵及Kappa系数的结果显示,六种方法中阴影建筑指数法的提取精度最高,总体精度为99.80%,Kappa系数为0.98。研究发现:OHS-2A卫星高光谱影像结合阴影建筑指数法,不仅精确提取了洱海这种大面积的水体信息,对山体环绕中的库塘水体也有很好的提取效果,而且还有效抑制了建筑物、山体阴影等环境噪声的干扰,非常适合高原湖泊的水体提取。
Taking the OHS-2 A hyperspectral satellite as the data source and Erhai lake of the Yunnan-Guizhou plateau as the main research area,a method extracting water bodies,called the shaded building index(SBI),is devised in the paper,which used the normalized difference water index and the strong absorption of the water in the near infrared band,and constructed the product relation between the both.This way can widen the gray difference between the water body and other objects,especially the building,and then extract the water body more accurately.The superiority of the method is showed by comparing with five existing water extraction methods:single-band threshold,spectral relation method,normalized difference water index,modified shadow water index,and comprehensive weight water index.The results of the confusion matrix and kappa coefficient show that the shaded building index method has a relatively high extraction accuracy among the above six different methods,with the overall accuracy of 99.80%and the Kappa coefficient of 0.98.It is found that the OHS-2 A hyperspectral image combined with the shaded building index method not only accurately extracts the large-area water information of Erhai lake,has a good extraction effect on the reservoir pond water body surrounded by mountains,but also effectively inhibits the interference of environmental noise such as architecture,mountain shadow.Thus the method is very suitable for water extraction of plateau lakes.
作者
尹慧新
樊彦国
YIN Huixin;FAN Yanguo(China University of Petroleum,Qingdao 266580,China)
出处
《航天返回与遥感》
CSCD
2020年第4期118-127,共10页
Spacecraft Recovery & Remote Sensing
关键词
阴影建筑指数
水体提取
高光谱影像
“珠海一号”卫星
遥感应用
shaded building index
water extraction
hyperspectral image
OHS-2A satellite
remote sensing application
作者简介
尹慧新,女,1996年生,2018年获山东农业工程学院测绘工程专业学士学位,现在中国石油大学(华东)测绘工程专业攻读硕士学位。研究方向为摄影测量与遥感。E-mail:yinhuixinxin@163.com。
