【摘要】 在遥感地学分析模型中,由于遥感信息模糊和不确定性的特点,其特征空间中的特征分布并不完全符合特定的高斯密度分布,而是分布形状各异,相互交错,用传统模型是很难获得特征的最优分布解的,而空间逐步寻优数据挖掘方法(SOMM)是在演化寻优理论的基础上,融合知识的参数化分布函数,来逐步分离特征空间,逐步降解的获得特征树状的层次结构。结合实例,用SOMM方法对遥感影像进行分类计算,并与传统的最大似然分类方法的分类结果进行了比较。更多还原

【Abstract】 In the model of the remote sensing and the analysis for Geo-science, for the reason of the ambiguity and indefiniteness of the remote sensing images, the distribution of the characteristic in characteristic space is not mostly accordance with the Gaoshi distribution. The distribution is in different forms and mutually crisscrossing. The traditional model is difficult to get optimization distribution for the characteristic. But, the method of Stepwise Optimization Making Modal for Spatial Data Mining (SOMM) is based on the evolution optimization theory and by mixing the parameters to stepwise separating so as to get the characteristic tree rank order structure. With the example the Remote sensing image are calculated using the SOMM and the maximum likelihood classifier. The results are compared.更多还原