摘要
无人机、卫星等获取的高分辨率遥感影像中不可避免地存在云层遮挡问题,这对遥感影像生产和应用造成一定程度的干扰。本文针对RGB彩色遥感影像中存在的云层遮挡问题,提出了一种基于RGB彩色遥感影像的快速云检测方法。首先根据RGB三通道的光谱信息生成粗云图,引入景观格局指数,然后排除地面非云层高亮图斑误差,利用原始影像作为引导,对初步云提取结果进行引导滤波,获得精细化云图,最后对结果进行矢量化处理,自动提取图斑的关键拐点,最大限度保留云的形状,获得真实、美观的矢量化成果。该方法对遥感影像的数据质量和波段数要求低,可以进行快速高精度的云检测。本文利用卫星影像和无人机影像进行快速云检测实验,证明了该方法的有效性。
Cloud cover is inevitable in high-resolution remote sensing images of UAV and satellite,which causes a certain degree of interference to the application of remote sensing images.Aiming at the problem of cloud occlusion in remote sensing image which only has RGB color.This paper proposes a fast cloud detection method based on remote sensing image of RGB.Firstly,the cloud image is generated according to the spectral information of RGB,and the landscape pattern index is introduced to eliminate the error of the highlight spot on the ground.Then,the original image is used as the guide to filter the preliminary cloud extraction results,and the refined cloud image is obtained.Finally,the results are vectorization automatically extract the key inflection points of the map patches,so as to retain the shape of the cloud to the greatest extent.This method has low requirements on the data quality and band number of remote sensing images,it can product fast and high-precision cloud detection.This paper uses satellite images and UAV images to carry out fast cloud detection experiments,which proves the effectiveness of this method.
作者
孙阳
郑新杰
SUN Yang;ZHENG Xinjie(Heilongjiang Institute of Geomatics Engineering,Harbin 150081,China)
出处
《测绘与空间地理信息》
2021年第S01期9-11,15,共4页
Geomatics & Spatial Information Technology
基金
国家重点研发计划项目(2016YFB0501401)资助。
关键词
RGB
遥感影像
云检测
引导滤波
RGB
remote sensing image
cloud detection
guided filtering
