【摘要】 以新发射的哨兵2号多光谱卫星影像为基础数据源,对城市水体信息进行了面向对象的提取研究。融合大气校正后的哨兵2号卫星10 m分辨率的可见近红外影像和20 m分辨率的短波影像,在多尺度分割的基础上,分析了影像对象的光谱和空间信息,建立了知识规则集;手动调节阈值,实现了水体信息的半自动提取。选取了城区和郊区两个研究区进行实验,总体精度均在95%以上,Kappa系数在0.75以上。结果表明,本文的水体提取方法取得了较好的效果,且城市主城区的水体提取精度高于郊区。更多还原

【Abstract】 In this paper, based on the newly launched Sentinel-2 multi-spectral instrument(MSI) imagery, we put forward a object-oriented urban water body extraction method. Firstly, we utilized four Near-Infra-Red(NIR) bands in 10 m resolution as a high-resolution image to sharpen the 20 m Short-Wavelength Infrared(SWIR) band after atmospheric correction. Then, we used the multiresolution segmentation algorithm to segment the combined images, and established the knowledge rule sets through the analysis of spectral and spatial information of image objects. Finally, we realized the semi-automatic extraction of water body by adjusting the threshold manually. We chose two study areas with different characteristics, and the total accuracies were above 95% and Kappa values were more than 0.75. The experiment results indicate that the proposed method has a better performance and higher accuracy in urban center region than suburbs.更多还原