论文部分内容阅读
空间聚类是空间数据挖掘中一个非常重要的方法.本文在分析 DBSCAN 算法不足的基础上,提出一种改进的空间聚类算法(AISCA).为了能够有效处理大规模空间数据库,算法采用一种新的抽样技术.另外,通过引入匹配邻域的概念,使得算法在聚类时不仅考虑空间属性也考虑非空间属性.二维空间数据测试结果表明算法是可行、有效的.
Spatial clustering is a very important method in spatial data mining.This paper proposes an improved Spatial Clustering Algorithm (AISCA) based on the analysis of the inadequate DBSCAN algorithm.In order to be able to deal effectively with large-scale spatial database, the algorithm uses a New sampling technique.In addition, by introducing the concept of matching neighborhood, the algorithm not only considers the spatial attribute but also considers the non-spatial attribute in the clustering.The test results of two-dimensional spatial data show that the algorithm is feasible and effective.