改进的基于DBSCAN的空间聚类算法研究

Research on Improving Clustering Algorithm Based on DBSCAN

针对DBSCAN算法I/O开销和内存消耗大的缺点,提出了基于层次合并的密度算法.该算法减少了DBSCAN算法中需要查询的点的数量,从而克服了DBSCAN算法I/O开销和内存消耗大的缺点.算法分析表明该算法对DBSCAN的改进是有效的.

To deal with the limitation of DBSCAN which I/O spending and memory expand is very big, the clustering algorithm based density and hierarchical is presented in this thesis. It gets over the limitation of DBSCAN by reducing the number of points that needed to be found. The analysis proves the new algorithm is effective.