面向GF-1影像的比值建筑用地指数构建

Construction of ratio build- up index for GF- 1 image

针对GF-1影像缺少短波红外波段的问题,根据建筑用地信息在缨帽变换(Kauth-Thomas transformation,KT)前2个分量具有高对比度的特征,提出了一种比值建筑用地指数(ratio build-up index,RBI)。在不同的时空条件下,利用该指数分别提取了湖北省钟祥市和襄阳市樊城区的建筑用地信息,并将其结果与常规分类器和建筑用地指数等方法提取的结果进行了比较。结果表明,RBI方法提取的建筑用地信息总精度和Kappa系数均较高,且对裸地信息具有较好的抑制能力。

In view of the facts that the high resolution image is usually lack of shortwave infrared wavelengths and that the first two components of KT transformation have the characteristics of high contrast,this paper proposes a ratio build- up index（ RBI） with GF- 1 images as the data source. Under different temporal- spatial conditions,the ratio build- up index was used to extract the build- up land information of Zhongxiang City and Fancheng District of Xiangyang City in Hubei Province,and the result was compared with results of the common classifiers and the common build- up index＇s extractions. It is shown that the accuracy and Kappa of RBI extraction are better,and RBI has good inhibition capability on bare land information.