基于影像增强和综合决策的VIRR影像云检测

摘要

To make full use of each VIRR band image,improving VIRR cloud detection effect,the cloud detection algorithm is studied based on confidence level of the sunny pixel.First,all kinds of remote sensing image of FY-3-VIRR multispectral data are enhanced by NDVI,RVI,NDWI,the water pixels are extracted,and image contrast is enhanced between the cloud pixel and the features pixel.Second,confidence level of the sunny pixel is determined,through the distribution of all kinds of pixels of each kind of enhanced images or single band images.Finally,a comprehensive decision was made,according to sunny pixels confidence level,the cloud mask is generated.By Compared with two kinds of results,the result of this method is better than that of directly using spectral information for cloud detection method.%为充分利用FY-3-VIRR(风云三号可见光红外扫描辐射计)VIRR各个波段的影像,提高VIRR云检测效果,提出基于像元晴朗置信水平的云检测算法.对VIRR多光谱遥感影像进行图像增强,包括NDVI,RVI,NDWI等,并对水像元进行提取,完成云像元和地物像元的对比增强;对每一种增强图像或者单波段图像中各类像元的分布进行分析,并确定像元的晴朗置信水平;综合决策对像元的晴朗置信水平进行确定,并生成云掩模.两种结果对比表明,本文所述方法云检测结果优于直接利用光谱信息云检测的方法.