时空对象的聚类方法与应用初探

Study on Clustering Methods and Applications for Spatio-temporal Objects

针对传统GIS数据模型描述信息有限以及对象化聚类分析内容不够全面的问题,提出了基于时空对象的聚类方法的流程和应用特点。首先总结了空间聚类和时空聚类的研究现状和主要方法;根据全空间信息系统中多粒度时空对象的描述框架及时空数据的"三维"特征,认为基于时空对象的聚类方法应包含3个方面:时空对象时间序列的相似性描述、基于时空对象的聚类计算及时空对象聚类的有效性评价;最后总结了该方法的特点并展望了其应用场景。基于时空对象的聚类方法研究有助于更全面地分析时空对象空间位置、属性特征及其变化特点,为多粒度时空对象的时空分析提供一种思路。

In view of the problem,the traditional GIS data models convey relatively limited information and the factors in clustering calculation are not comprehensive,the process and application characteristics of clustering analysis based on spatiotemporal objects are proposed.Firstly,main research and methods of spatial clustering and spatio-temporal clustering are summarized.On the basis of description framework of multi-granularity spatio-temporal objects in pan-spatial information system,three aspects in ST object clustering are concluded:similarity description of time series,clustering calculation based on ST objects,and validity assessment of clustering.Finally,we summed up the advantages and assumed the application scenarios.Clustering method based on ST objects contributes to comprehensive analysis of its spatial position,attribute character and variation.It also provides a ST analysis method for multi-granularity spatio-temporal objects.