关键词: 空间聚类, K均值, 空间关系, 方向关系

Abstract: Under the bigger data conditions, the traditional spatial relationships describe only two space objects and show many problems of redundant data storage, retrieval and so on. Firstly, the method uses clustering algorithm for clustering space objects, and then indicates the direction of relationship between space objects which are based on the clustering results. Thus this paper proposes clustering algorithm based on K-means and density, and indicates the method of combination of the direction relationship and clustering algorithm. The methods enhance the ability of handling massive spatial data in spatial database, save a lot of storage space and improve the query speed of data.

Key words: clustering algorithm, K-means, spatial relationship, direction relationship