高分六号影像不透水面提取研究

Study on extraction of impervious surface based on GF-6 image

针对高空间分辨率卫星影像空间分辨率高、波段较少这一特点,该文以高分六号多光谱影像为数据源,通过分析各类地物的光谱特征曲线,建立了一种新指数。该指数通过增强植被和裸土信息,抑制不透水面信息来扩大不透水面和植被、裸土的差异,实现对不透水面的分离提取。验证阶段将该指数用于天津市河西区和河北省易县中心城区,并将提取结果与垂直不透水面指数(PISI)、比值居民地指数(RRI)的提取结果进行对比。结果表明:该方法在河西区和易县的提取效果都要优于PISI和RRI,并且两块实验区提取结果的总体精度分别为94.0%和91.7%,具有一定的稳定性。

Aiming at the characteristics of high spatial resolution satellite images with high spatial resolution and fewer bands,this paper took GF-6 multi-spectral image as the data source and established a new index that could extract impervious surface by analyzing the spectral characteristic curves of various ground objects.By enhancing the information of vegetation and bare soil,suppressing the information of impervious surface,the index could enlarge the difference of impervious surface,vegetation and bare soil,and realize the separation and extraction of impervious surface.In the validation stage,this index was applied in Hexi District of Tianjin City and the central urban area of Yi County in Hebei Province,and the extraction results were compared with the extraction results of PISI and RRI.The results showed that the extraction efficiency of the proposed method in Hexi District and Yi County was better than that of PISI and RRI,and the overall accuracy of the extraction results in two experimental areas was 94.0%and 91.7%respectively,which had certain stability.