Abstract: Spatial clustering has been a major research field in spatial data mining;it aims to discover some useful patterns or outliers in a spatial database.In practice,spatial obstacles,as river or mountains should be fully considered in the process of spatial clustering.On that account,a novel spatial clustering method considering spatial obstacles is proposed in this paper.Delaunay triangulation is employed to model spatial proximate relations among entities,and the method can automatically discover clusters with complex structures without user-specified parameters.Experiments on both simulated database and real-world database are utilized to demonstrate the effectiveness and advantage of our method.