CURD is one type of clustering algorithm based on reference point and density. This algorithm is similar to DBSCAN in processing arbitrary shape clustering ability and has linear time complexity of K-MEANS algorithm. CURD algorithm needs to set Radius and t, so the whole process of clustering needs manual intervention which has the similarity with most of clustering algorithm. This paper proposes SA-CURD clustering algorithm based on CURD which can automatically set Radius and t by analyzing

more »

... aset statistics, in order avoid manual intervention in the process of clustering and achieve complete automation. Experiments show that SA-CURD can rationally choose Radius and t and get highly precise clustering results.