Computer Engineering and Applications ›› 2010, Vol. 46 ›› Issue (20): 133-137.DOI: 10.3778/j.issn.1002-8331.2010.20.038

• 人工智能 • Previous Articles     Next Articles

Study of boundary detecting technique with clustering

QIU Bao-zhi,JU Chang-tao   

  1. School of Information & Engineering,Zhengzhou University,Zhengzhou 450001,China
  • Received:2010-04-14 Revised:2010-05-17 Online:2010-07-11 Published:2010-07-11
  • Contact: QIU Bao-zhi

具有聚类功能的边界检测技术的研究

邱保志,琚长涛   

  1. 郑州大学 信息工程学院,郑州 450001
  • 通讯作者: 邱保志

Abstract: In order to detect boundary points of clusters quickly and efficiently,this paper proposes an algorithm named cluster boundary detecting algorithm based on Delaunay triangulation(DTBOUND).DTBOUND divides the dataset into internal set and external set by calculating the coefficient of variation of each data point of the triangulation.Then,the depth-first traversal is made from each of the unclassified internal point to put the internal points connected with each other and external points connected with the internal points into a cluster.Finally,boundary points are extracted from the cluster obtained before.There is only one parameter in DTBOUND.Experimental results show that DTBOUND can quickly and effectively identify cluster boundary points and clusters of arbitrary shapes,different sizes and different densities.

Key words: boundary point, clusters, Delaunay triangulation, coefficient of variation

摘要: 为快速有效地检测聚类的边界点,提出了一种新的基于三角剖分的聚类边界检测算法DTBOUND。该算法通过计算三角剖分图中每个数据点的变异系数将数据集分解成内部点和外部点两部分,然后从每一个未分类的内部点开始进行深度优先遍历,将相连的内部点以及和内部点相连的外部点作为一个聚类;最后从得到的聚类中提取边界点。该算法只有一个参数(变异系数阈值β),实验结果表明该算法可以快速、有效地识别任意形状、不同大小和不同密度的聚类和聚类的边界点。

关键词: 边界点, 聚类, 三角剖分, 变异系数

CLC Number: