论文部分内容阅读
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．此方法成功地用于主扫描图象分类
Different types of ground objects have different eigenvectors in the multidimensional spectral space due to different radiation spectral distributions. These vectors can be characterized by binary numbers. The dynamic clustering method introduced in this paper is an unsupervised classification method based on binary eigenvectors. This method has been successfully used for main scanning image classification