空间知识挖掘的自然面群聚集度聚类方法

空间聚类是挖掘空间知识的重要手段之一。针对现有方法难以处理几何、分布特征差异大的面群聚类问题,本文提出了一种面要素分布密度的描述参数—聚集度,并设计了一种自然面群聚类方法。首先,分析了面要素分布密度的影响因子,定义了聚集度的概念,设计其计算方法并验证其有效性及优势;然后,基于聚集度和边界最短距离建立相邻面从属关系,识别聚类中心,完成初始群组的构建;最后,围绕群组特征设计了边缘检测和群组合并模型,实现了邻近相似群组的合并。试验表明,相较于最小生成树、强度函数聚类方法,本文方法兼顾几何特征、分布特征的复杂性,有效提升了自然面群的聚类效果。

An aggregation index clustering method of natural polygon features for spatial knowledge mining

Spatial clustering is one of the important methods to mining spatial knowledge.Existing methods fail to cluster natural polygon features with great differences in geometry and distribution.Hence an aggregation index is proposed to measure distribution density,and a new natural polygon feature clustering method is designed.First,the formula of aggregation index is designed and its effectiveness is verified.Then,on basis of aggregation index and the shortest distance,the affiliation relationship of adjacent polygon features is established to identify the clustering center.Thus,initial clustering group is constructed.Finally,border feature detection principle and adjacent group merging model are provided to obtain better clustering results.Experiments show that compared with MST and MSSCP,the method proposed can take the complexity of geometric and distribution characteristics into account and effectively improve the clustering results of natural polygon features.