多层次空间同位模式自适应挖掘方法

发布时间：2018-02-28 00:04

本文关键词： 空间异质性空间同位模式自适应聚类叠置分析　出处：《测绘学报》2016年04期 　论文类型：期刊论文

【摘要】：空间同位模式挖掘旨在从空间数据中发现频繁发生在邻近位置的事件集合,对于揭示地理现象间的共生规律具有重要价值。由于地理现象的空间异质特质,空间同位模式也存在区域性分异的特点,在不同空间层次上的分析结果各异。然而,现有方法仅从全局视角挖掘空间同位模式,发现局部空间同位模式依然是一个亟待解决的难题。为此,本文基于由整体到局部的思想,提出了一种多层次空间同位模式自适应挖掘方法。首先,从全局视角提取频繁的空间同位模式,将全局不频繁的空间同位模式作为候选的局部空间同位模式;然后,通过对候选局部同位模式进行自适应聚类自动识别其局部分布区域,并在这些局部区域内度量候选模式的频繁程度;进而,提出了一种叠置推绎的方法,从频繁子模式的局部区域中进一步推绎获得超模式的局部分布区域,最终生成所有频繁的局部空间同位模式集合。通过试验分析与比较发现,本文方法不仅可以发现全局的空间同位模式,还能有效提取具有区域性分布特征的局部空间同位模式,可以从多个空间层次上反映地理事件间的共生规则。
[Abstract]:The spatial co location pattern mining is to find a collection of data from space occurred frequently in the proximity of the event, is of great value to reveal the symbiosis of geographical phenomena. The spatial heterogeneity characteristics due to the phenomena of geography, spatial co location pattern also has the characteristics of regional difference, the analysis results in different spatial levels were different. However, the existing only the method of mining spatial co location pattern from the global perspective, local spatial co location pattern is still a problem to be solved urgently. Therefore, this article from the whole to the part based on the idea, puts forward a method of multi-level adaptive spatial co location pattern mining. Firstly, extract the spatial co location pattern frequently from the global perspective, will serve as the local space parity mode candidate spatial co location patterns global frequent; then, through adaptive clustering for pattern recognition of the candidate local parity Local distribution, frequent degree and measure candidate patterns in these areas; and then, puts forward a method of overlapping scaling, further pushing the local distribution area of super mode from the local frequent pattern in the final set of all frequent local spatial co location patterns. Through analysis and test in comparison, spatial co location patterns this method not only can be found globally, can effectively extract the local spatial co location pattern with the distribution characteristics of the region, can reflect the common rules among geographical events from multiple spatial levels.

【作者单位】：中南大学地球科学与信息物理学院地理信息系;
【基金】：湖南省自然科学杰出青年基金(14JJ1007) 国家自然科学基金(41471385) 资源与环境信息系统国家重点实验室开放基金~~
【分类号】：P208

【相似文献】