一种有效聚类算法的研究和实现

Research and realization of an efficient clustering algorithm

提出了一个基于数学形态学的三维空间聚类算法。该算法通过闭合运算,将空间对象聚成类,一次完成三维空间聚类,可以快速处理非凸的、复杂的聚类形状。由于该算法基于数学形态学,所以易于实现其高性能并行算法。采用实例将算法与普通聚类算法进行了性能比较。

Based on mathematical morphology, a new algorithm of 3D spatial clustering was presented, which clustered spatial objects by closure operation. This algorithm could not only complete 3D spatial clustering at a time, and process clustering in-convex and complicated objects rapidly. On the basis of mathematical morphology, its high performance parallel algorithm was easy to realize. Experiments show that the algorithm is better than general clustering algorithms in some cases.