论文部分内容阅读
迭代自组逐步聚类法(ISOMIX)是一种新的数理统计方法。此法在美国已用于对陆地卫星数据的分析,用电子计算机进行图象识别,属于非监督分类。非监督分类也就是无先验分类。此法特点是:在运用电子计算机进行自动识别分类之前,不需要事先选择一些地面标志分类,对计算机进行训练和控制,而是通过逐步聚类直接建立决策规则。这种方法的理论依据是:各类样本具有内在的相似性,把各样本在空间分布按相似性分割或合并成集群,再
Iterative self-organizing stepwise clustering (ISOMIX) is a new mathematical statistics method. This method has been used in the United States for the analysis of terrestrial satellite data, using computer graphics for image recognition, and is unsupervised. Unsupervised classification is no prior classification. This method is characterized by: before the use of electronic computers for automatic identification classification, do not need to select some of the land mark classification in advance, the computer training and control, but through the gradual clustering directly establish decision rules. The theoretical basis of this method is that all kinds of samples have inherent similarities, and the samples are divided or merged into clusters according to similarity in spatial distribution, and then