【作者】 刘启亮； 邓敏； 李光强； 王佳璆； 彭东亮；

【Author】 LIU Qiliang,DENG Min,LI Guangqiang,WANG Jiaqiu,PENG Dongliang (Department of Surveying and Geo-informatics, Central South University, Changsha 410083, China)

【机构】 中南大学测绘与国土信息工程系；

【摘要】 空间异常探测是空间数据挖掘的一种重要手段,旨在发现小部分偏离普遍模式的异常实体,对揭示地理现象或地理过程潜在的发展变化规律有着重要意义。现有的空间异常探测方法仅能在单一尺度上探测异常,而现实中空间异常实体(或现象)通常在不同的尺度出现,从而导致现有的探测方法不能完全探测出隐藏的空间异常。并且现有的探测方法中空间异常度量指标受空间邻近域的选择及空间邻近域内异常值的影响较大,直接影响了空间异常探测结果的可靠性。为此,本文提出了基于层次聚类的多尺度空间异常探测方法,并给出了一种稳健的空间异常度量指标,从而能够更准确地探测不同尺度下的空间异常。通过实例验证了本方法的可行性和正确性。更多还原

【Abstract】 Spatial outlier detection has been an important approach for spatial data mining and knowledge discovery. Spatial outliers are the entities whose non-spatial attributes are significantly different from the value of the entities in their spatial neighborhoods. Indeed, spatial outliers may be used to discover potential geography phenomenon or geographic change process. Most current spatial outlier detection methods are mainly focused on the detection of spatial outliers only with single scale, thus difficult for the detection of multi-scales spatial outliers. At the same time, the entities could be potentially wrongly identified as spatial outliers when there are several real outliers in their spatial neighborhoods. In addition, the dynamic construction methods of spatial neighborhood (e.g.K-NN) usually obtain different results with regards to different parameters, which may puzzle the operators. In order to overcome such limitations, an approach for multi-scale spatial outlier detection method based on hierarchical clustering is proposed in this paper, where a robust spatial outlier measurement (RSOM for short) is utilized to determine the spatial outliers. This method is able to detect spatial outliers with multi-scales accurately. A practical example is employed to demonstrate the validity of the spatial outlier detection method proposed in this paper.更多还原