一种主扫描图象的二进制特征向量动态聚类方法  

AN UNSUPERVISED DYNAMIC CLUSTERING ALGORITHM FOR LANDSAT THEMATIC MAPPER BASED ON A VECTOR OF BINARY FEATURES

作　　者：安斌[1] 陈书海[1] 张平[1] 严卫东[1] 

机构地区：[1]西北核技术研究所

出　　处：《光子学报》1999年第5期473-477,共5页Acta Photonica Sinica

摘　　要：不同类型的地物，由于辐射光谱分布不同，在多维光谱空间中构成不同的特征向量，这些向量可以用二进制数码表征．本文介绍的动态聚类方法便是一种基于二进制特征向量的非监督分类方法．Different types of terrain would lie on the different site in the spectral space,which correspond to a vector of binary spectral features.The reflecting spectral of terrain could be described conveniently and exactly with the binary vector presented by Mark.J.Carlotto 1 in 1996.This vector consists of several binary codes,which result from relative value between bands.In this paper,a unsupervised clustering algorithm for Landsat thematic mapper based on a vector of binary spectral features is presented,The algorithm could be used to cluster TM data successfully compared with K mean clustering algorithm.

关 键 词：主扫描图象 二进制特征向量 动态聚类 图象 

分 类 号：TN919.8[电子电信—通信与信息系统]