Abstract: In order to solve Reverse Nearest Neighbor(RNN) search in a dataset, Voronoi diagram and the hulls of dataset are used for RNN search. By determinating the location relation between query point and convex hulls, many points can be cut. And an algorithm and judgment method for the change of the RNN of query point are given when data points are added or deleted. To be convenient to search, corresponding spatial storage structures are designed. Comparative analysis shows that this method has evident advantages when many query points are dealt with.

Key words: Reverse Nearest Neighbor(RNN), Voronoi diagram, convex hull

摘要： 为了解决反向最近邻查询问题，利用Voronoi图及数据集中点的凸包进行反向最近邻查询，通过判断查询点与凸包的位置关系，可去除大量的数据点，并且给出在数据点被加入或删除后，对查询点的反向最近邻变化情况的判断方法与算法。为了便于查询，设计相应的空间存储数据结构。比较分析表明，该方法在处理多个查询点的反向最近邻时有一定的优势。

关键词: 反向最近邻, Voronoi图, 凸包

CLC Number: 

• TP311.13