【摘要】 传统的Hilbert Packed R-树是利用Hilbert值对空间实体依次进行压缩,算法简单快速,然而空间位置上邻近的空间实体的Hilbert值并不一定相邻,使得在数据分布不均匀时,查询效率开始下降;递归聚类的算法虽然解决了以上问题,但是它计算复杂,而且容易造成R-树的不平衡,以至降低了存储利用率和检索的效率。文中对两种方法加以综合,提出了一种新的批量加载R-树的算法—HilCluster。实验结果表明,新算法不仅继承了Hilbert Packed R-树构造过程时间消耗低、存储利用率高的优点,还使得查询效率进一步提高。更多还原

【Abstract】 Traditional Hilbert Packed R-Tree packs the spatial objects by the turn of their Hilbert values.Though the algorithm is simple and rapid,the spatial objects which neighboring in their spatial location do not necessarily neighboring in their Hilbert values.Because of this,its query efficiency decline for the data distributed unevenly.The algorithm clustering recursively resolved the problem,but it has high construction expense,and the R-Tree constructed by means of it usually is imbalance,which result in low space utilization and efficiency.In this paper,united above two methods,proposed a new bulk-loading algorithm for the R-Tree called HilCluster.The experimental data indicates that the new algorithm not only inherit Hilbert Packed R-Tree’s low construction expense and high space utilization percent,but also has better performance in searching.更多还原