空间聚类是空间数据挖掘的重要手段之一.本文研究了一种基于质心点距离的Max-min distance空间聚类算法:通过加载园地图斑数据,计算其园地图斑质心,判断聚类中心之间的距离,并将符合条件的园地图斑进行聚类,最终将聚类结果可视化表达.本文的算法是利用Visual Studio 2017实验平台和ArcGIS Engine组件式开发环境,采用C#语言进行编写.实验结果表明:1)Max-min distance聚类通过启发式的选择簇中心,克服了K-means选择簇中心过于邻近的缺点,能够适应嵩口镇等山区丘陵地区空间分布呈破碎的园地数据集分布,有效地实现园地的合理聚类;2)根据连片面积将园地空间聚类结果分为大中小三类,未来嵩口镇可以重点发展园地连片规模较大的村庄,形成规模化的青梅种植园.
Garden Spatial Clustering Based on Max-min Distance Clustering Algorithm:Taking Songkou Town, Yongtai County as an Example
Spatial clustering is one of the important means of spatial data mining. In this paper, a Max-min distance spatial clustering algorithm based on centroid point distance is studied: by loading the garden spot data, calculating the centroid of the garden spot, jud-ging the distance between the cluster centers, and classifying the eligible garden spots, it performs clustering and finally visualizes the clustering results. The algorithm in this paper is programmed in C# language using Visual Studio 2017 as the experimental platform and ArcGIS Engine component development environment. The experimental results show that: 1) Max-min distance clustering im-proves the shortcomings of K-means selection of cluster centers that are too close by heuristically selecting cluster centers, and can a-dapt to the broken spatial distribution in mountainous and hilly areas such as Songkou town. The distribution of the garden data set can effectively realize the reasonable clustering of the garden; (2) According to the contiguous area, the spatial clustering results of the garden are divided into three categories: large, medium and small. The villages of large scale contiguous area in Songkou town can be developed to form a large-scale green plum plantation.
Max-min distance clustering algorithmgardenGISSongkou Town
万方数据
