遥感影像分类的归类学习方法  

Remote Sensing Image Classification Based on Inductive Learning

作　　者：孙立新 罗高平 张怡梅[3] 

机构地区：[1]哈尔滨工程高等专科学校,哈尔滨市中山路153号150008 [2]深圳市龙岗区城建开发实业总公司,深圳市龙岗区中心城龙城大道518172 [3]哈尔滨师范大学,哈尔滨市和兴路50号150080

出　　处：《测绘工程》1998年第3期39-43,49,共6页Engineering of Surveying and Mapping

摘　　要：遥感影像分类专家系统是遥感分类研究中的一个重要发展方向，然而，传统的统计模式识别法和人工神经网络分类法除了能完成具体的影像分类外，不能提供易于被人类理解的分类知识。文中介绍一种基于扩张矩阵的示例学习方法，并将其应用于遥感影像分类知识的自动获取。Remote sensing image classification expert system is one of the important developing directions of remote sensing classification research. Conventional classification approaches, based on pattern recognition and artificial neural network, can only fulfill the particular classification tasks, instead provide classification knowledge which is easy to be understood by human. In this paper, an inductive learning method based on extension matrix is introduced and applied to the automatic acquistition of classification knowledge for remotely sensed images.

关 键 词：归纳学习 扩张矩阵 遥感 影像分类 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]