基于数据分区的并行DBSCAN算法

DBSCAN是基于密度的聚类算法的一个典型代表算法，它对空间数据库聚类有很好的性能．然而，在对大规模数据库聚类时，DBSCAN需要大量内存支持并伴随着I／O开销．随着高性能计算机的发展，特别是集群式计算机出现，给我们提供了一种解决DBSCAN算法缺陷的方法，本文提出一种建立在集群式高性能计算机上基于数据分区并行DBSCAN算法．测试结果表明，它极大地降低了DBSCAN对时间和空间的需要．

Data-Partitioning-Based Parellel DBSCAN Algorithm

As an outstanding representative of clustering algorithms, DBSCAN algorithm show good performance in spatial data clustering. However, for large spatial databases, original DBSCAN requires large volume of memory support and could incur substantial I/O costs, with the rapid development of high performance computer, especially appearance of the cluster computers give a means to overcome the defect of original DBSCAN algorithm. In this article, we will propose a based-on cluster computering parallel DBSCAN algorthim. It is also a data-partitioning-based DBSCAN algorithm. Experimental results show that the new agrithom is superior to the original DBSCAN in efficiency.