多光谱卫星图像的一种模糊聚类方法

基金项目: 参作气字2002第35号基金资助

A FUZZY CLUSTERING METHOD FOR MULTI-SPECTRAL SATELLITE IMAGES

1.
Institute of Meteorology PLA University of Science and Technology, Nanjing 211101, China
2.
Institute of Meteorology PLA University of Science and Technology, Nanjing 211101, China;Nanjing University of Information Science & Technology, Nanjing 210044, China
3.
Nanjing University of Information Science & Technology, Nanjing 210044, China

摘要: 基于二维光谱特征空间,用模糊C均值(FCM)聚类方法,对多光谱静止卫星(GMS-5)图像进行了云分类试验,得到了比较合理的分类结果。该方法利用不同光谱通道的卫星云图光谱特征构造出一个二维光谱特征空间,对云图在特征空间上的光谱特征点进行FCM聚类,然后与已知云类样本的特征进行比较,确定出各聚类域的类属,进而得到二维光谱空间的云分类图,实况接收的云图可通过查验特征像素点在分类图中的落区位置来实现云的分类。
- FCM /
- 卫星图像 /
- 云分类
Abstract: Based on the two-dimensional spectral feature space combining with fuzzy C-mean clustering(FCM), experiment of cloud classification were carried out using GMS images and the result was reasonable. A two-dimensional spectral feature space was constructed, in which the image pixels were classified to several cluster groups by FCM. By extracting the characteristics of the cloud and comparing the cloud images with the weather fact, types of each cluster group were determined. These groups are made up of classification charts that are built on the two-dimensional spectral space. According to the charts, the type of the pixels in original cloud images can be made certain through finding out which group the pixels fit in the charts.
- FCM /
- satellite image /
- cloud classification

[1]	KOFFLER, DECOTIIS R A G, RAO P K. A procedure for estimating cloud amount and height from satellite infrared radiation data[J]. Mon Wea Rev, 1973,101: 240-243.
[2]	DESBOIS M, SEZE G, SZEJWACH G. Automatic classification of clouds on METEOSAT imagery application to highlevel clouds[J].JAppl Meteor, 1982, 21: 401-402.
[3]	WELCH R M, NAVAR M S, SENGUPTA S K.The effect off resolution upon texture-based cloud field classification[J].J Geophys Res, 1989, 94: 14767-14781.
[4]	陈子通.卫星图像像元分析的方案研究和初步试验[J].热带气象学报,2003,19:177-183.
[5]	郁凡.多光谱GMS卫星图像气象特征量的提取及其在中尺度数值预报模式中的应用[D].南京:南京大学,1998,10.
[6]	BEZDEK J C.A convergence theorem for the fuzzy ISODATA clustering algorithms[J].IEEE Tran,1980,PAMI-2:1-7.