关键词: 空间数据, R-树, 空间索引, 聚类算法

Abstract: The current R-tree spatial clustering algorithms use a predefined k-value for clustering, and choose the initial clustering centers arbitrarily. The clustering results are easily dominated by the initial k-value and the outlier data. In order to solve these problems, this paper proposes a novel spatial clustering algorithm called DKSC (dynamical k-value spatial clustering algorithm), which dynamically determines the optimized k-value when constructing the tree. By choosing an optimized k-value, the algorithm allows the spatial data in the same subspace to be organized into the same sub-tree, and builds an efficient R-tree layer by layer from the root to leaves. The experiments conducted with real and simulated datasets show that the proposed algorithm can build optimized R-tree spatial indexes and improve retrieval efficiency.

Key words: spatial data, R-tree, spatial index, clustering algorithm