空间和属性双重约束下的自组织空间聚类研究  

Self-organizing Spatial Clustering Under Spatial and Attribute Constraints

作　　者：焦利民[1,2] 洪晓峰[1] 刘耀林[1,2] 

机构地区：[1]武汉大学资源与环境科学学院,武汉市珞喻路129号430079 [2]武汉大学地理信息系统教育部重点实验室,武汉市珞喻路129号430079

出　　处：《武汉大学学报（信息科学版）》2011年第7期862-866,共5页Geomatics and Information Science of Wuhan University

基　　金：国家自然科学基金资助项目(40901188);地理空间信息工程国家测绘局重点实验室开放研究基金资助项目(200906);中央高校基本科研业务费专项基金资助项目(4082002)

摘　　要：形式化定义了双重聚类的聚类准则及其判定方法,提出了双重聚类的两步法求解思路和自组织双重聚类算法。通过实例验证了该算法的可行性,自组织双重聚类可以发现非空间属性的聚集、延伸等空间分布特征,可以发现任意复杂形状的聚类,并降低了人为影响。Spatial clustering under spatial and attribute constraints is the clustering analysis on the spatial dataset with non-spatial attributes,which is named dual clustering.The result of dual clustering should be spatially continuous and attributively aggregative.The essence of dual clustering is to find out the clustering and distribution rules of non-spatial attributes.This paper presents the formalized definition of dual clustering,proposes the two-step strategy and self-organizing dual clustering algorithm.Case study verifies the algorithm and shows that the self-organizing dual spatial clustering can find the spatial distribution rules of non-spatial attributes,such as clustering and stretching.Self-organizing dual clustering can detect clusters with complicated shape and reduces the artificial influence.

关 键 词：空间聚类 空间和属性约束 双重聚类 自组织网络 属性距离 

分 类 号：P208[天文地球—地图制图学与地理信息工程] P273[天文地球—测绘科学与技术]