基于SPOT卫星影像的去模糊处理
杨晓峰1 ，郑有飞1 ，温兴平2 ，张育萍3
1.南京信息工程大学环境科学与工程学院，南京210044； 2.中国地质大学数学地质遥感地质研究所，武汉430074； 3. 江苏省张
家港市气象局，张家港215600
YANG Xiao-feng 1 , ZHENG You-fei 1 , WEN Xing-ping 2 , ZHANG Yu-ping 3
1. Institute of Environmental Science and Engineering, Nanjing University of Information Science and Technology,
Nanjing 210044, China；2. Institute of Mathematic Geology and Remote Sensing Geology, China University of
Geosciences; Wuhan 430074, China；3．Zhangjiagang Meteorological Bureau of Jiangsu Province, Zhangjiagang 215600,
China
摘要 用模糊区|清晰区平均反射率匹配法对SPOT 5卫星影像进行去模糊处理。首先，结合SPOT 5各波段的统计特征，从影像上识别
出模糊区； 然后，对影像进行非监督分类，对同一分类区内像元在模糊区和清晰区的辐射值进行比较，计算出气溶胶空间分布； 最
后，将气溶胶分布输入MODTRAN模型中，对影像进行大气校正。研究中将反演的气溶胶浓度分布重采样到全色波段的分辨率，用
MODTRAN对全色波段进行大气校正，也取得了较好效果。研究得出模糊区|清晰区平均反射率匹配法是一种有效的去模糊处理方法，且
可将处理后的影像转换得出反射率，便于不同时相遥感影像进行对比。
关键词 ：
重点污染源 ,
遥感解译 ,
城市环境
Abstract ：This paper deals with the methods for haze removal based on SPOT5 remote sensing image by using haze-
clear areas reflectance matching. The procedure is as follows: First, the haze area is recognized by using band
statistic values; second, the remote sensing image is classified by using unsupervised classification, then the
aerosol optical depth distribution is calculated by comparing the radiation values in haze and clear areas and,
finally, the image can be atmospherically corrected by using MODTRAN model. The panchromatic image can be
atmospherically corrected in a similar way by using the resample aerosol optical depth distribution. It is concluded
that the haze-clear area reflectance matching is an effective haze removal method. In addition, the image can be
transformed to reflectance and compared with different remote sensing images.
Key words ：
Main pollution sources
Remote sensing interpretation
Urban environment
收稿日期: 2007-04-10
出版日期: 2009-07-13
通讯作者:
杨晓峰（1973-），女，硕士研究生，工程师，研究方向为环境气象和环境遥感。
引用本文:
杨晓峰, 郑有飞, 温兴平, 张育萍. 基于SPOT卫星影像的去模糊处理[J]. 国土资源遥感, 2008, 20(1): 31-33.
链接本文:
https://www.gtzyyg.com/CN/10.6046/gtzyyg.2008.01.06
或
https://www.gtzyyg.com/CN/Y2008/V20/I1/31
Viewed
Full text
Abstract
Cited
Shared
Discussed
2008, Vol. 20 
