一种提取城市多种不透水层的垂直不透水层指数

针对中低分辨率影像中不透水层的异质性及其与裸土光谱的易混性两类问题,选用蓝、近红外波段进行线性组合,构建了一种新的不透水层提取指数——垂直不透水层指数(PII)。该指数考虑了不透水层和其他地物在光谱空间的差异与不透水层的内部异质性,并以"不透水层线"与"土壤线"夹角的角平分线作为PII的参照线,实现了自适应的不透水层提取。本文将PII指数应用于武汉和北京不同场景中,并对比归一化建筑物指数(NDBI)、比值居民地指数(RRI)以及生物物理组份指数(BCI)的提取结果。试验表明:1在裸土较少、地形平坦的武汉市区域和裸土较多、地形起伏的北京市区域,PII指数均能有效减弱裸土的混淆影响,不透水层提取精度分别达到96.05%和96.76%,优于其他3种指数;2PII指数不仅增强了不透水层与其他地物的可区分性,还保持了不透水层类内的相似性,在城市不同场景中的不透水层提取精度均能达到90%以上。由于PII指数是一种线性组合形式的指数,能够根据研究区的地物光谱自适应调整指数的方程系数,从而能适用于不同研究区,在裸地较多的地区优势尤为明显。

Perpendicular Impervious Index for Remote Sensing of Multiple Impervious Surface Extraction in Cities

Focusing on the issues of impervious layers' heterogeneity and confusion with soil, a method is presented-perpendicular impervious index (PII), which considers blue and near infrared bands selected based on spectral characteristics of ground objects in LandSat-8 images.The PII is established in a linear form, and its reference line is calculated based on the angle bisector of impervious layer line and soil line.Impervious surfaces are extracted using PII on LandSat-8 images of Beijing and Wuhan, which is compared with the normalized difference building index (NDBI), ratio resident-area index (RRI) and biophysical composition index (BCI) in the same areas.The conclusions are as follow: ① PII is superior than other indexes in separating impervious layers from bare soil in both the Wuhan and Beijing, the extracting accuracy is 96.05% and 96.76%,respectively;② PII is also effective in different environments, where the impervious layer shows various spectrums.Due to the linear combination form, PII can adjust its coefficients depending on spectra of ground objects in different study areas, which gets higher accuracy in extracting impervious layers than other indexes,especially in regions containing more bare soil.