中文摘要：

针对R-树索引空间查询效率低下的问题，提出一种基于节点分裂优化的R-树索引结构：SR-树索引。SR-树索引在节点分裂过程中，通过增加叶子节点的空间数据聚集性来减少叶子节点最小外接矩形的覆盖面积。为了有效降低磁盘读写消耗，SR-树节点在写入索引时，首先将索引树在内存中建好，然后在文件中写入树信息，最后通过递归的方式写入节点。实验结果表明，与R-树索引相比，SR-树索引可以在减少最小外接矩形重叠面积的同时，有效降低查询响应时间，从而达到提高查询效率的目的。

英文摘要：

Aiming at the problem of low spatial query efficiency in R-tree index, this paper presented SR-tree, an R-tree variant index structure that was based on node splitting optimization. During the process of node splitting, SR-tree improved spatial data aggregation of leaf nodes in order to reduce the overlapping area of minimum bounding rectangle （MBR） in leaf nodes. For the sake of decreasing disk input and output consumption, the nodes of SR-tree were written to the index as follows. First, index tree was built in the memory beforehand. Second, the information of tree was written into files. At last, the nodes were written recursively into the tree. The experiments demonstrate that SR-tree can reduce MBR overlapping areas and can decrease query response time, which in turn achieves the purpose of improving query efficiency.