【摘要】 云是光学卫星遥感影像主要的噪声源之一,会遮蔽或不同程度干扰影像上的地物信息,影像云检测得到的云量和云掩膜对卫星影像的生产和后续的数据分析使用非常重要。文章针对"高分七号"(GF-7)卫星等高分辨率遥感影像的波段特点,提出了一种基于光谱信息和自适应阈值的4波段多光谱影像自动云检测方法。该方法首先对待检测影像进行分块并计算其云指数,然后基于云指数影像上云和地物的差异特性自适应分割得到云区,最后对分块影像拼接并进行形态学处理,得到最终的云量和云掩模。GF-7卫星多光谱影像自动云检测实际应用结果显示,厚云和半透明的薄云识别精度较高,可以比较精确地检测影像上不同类型的云,能够满足海量数据自动化云检测处理的需要。更多还原

【Abstract】 Cloud is one of the main noise sources of optical satellite remote sensing image,which would block or interfere with the ground feature information to different degree,the cloud amount and cloud mask obtained from cloud image detection are very important for the production of satellite imagery and subsequent data analysis.Aiming at the band characteristics of high resolution remote sensing images such as GF-7 satellite,this paper proposes an automatic cloud detection method for 4-band multispectral images based on spectral information and adaptive threshold.This method first divides the to-be-detected image into blocks and calculates its cloud index,and then adaptively segments the cloud area based on the difference characteristics of clouds and ground features on the cloud index image,and finally stitches the block images and performs morphological processing to obtain the final cloud amount and cloud mask.This method is used for automatic cloud detection of GF-7 satellite multispectral image.The results show that the method can detect different types of clouds on the image more accurately,especially with high recognition accuracy for thick clouds and semi-transparent thin clouds,thus meeting the needs of automated cloud detection processing of massive data.更多还原