一种极大中心间隔的核可能性C均值聚类算法

Kernel possibilistic C-means clustering algorithm based on maximum center interval

传统核可能性C均值(KPCM)算法仅考虑类内的紧密性而忽略了类间的距离关系,在对边界模糊的数据进行聚类分析时,会引起因聚类中心距离小或重合引起的边界点误分问题。为解决上述问题,在核可能性C均值基础上引入高维特征空间中的类间极大惩罚项和调控因子λ,构造了全新的目标函数,称为极大中心间隔的核可能性C均值(MKPCM)聚类算法。该算法通过类间极大惩罚项使类间距离极大化,并利用调控因子λ合理控制类间距,较好地避免了类中心间距离小或重合的现象。通过大量的实验证明,算法对于边界模糊的数据聚类效果优于传统的聚类算法;在图像分割的实际应用中,算法也明显优于传统的聚类算法。

The traditional Kernel Possibilictic C-Means(KPCM)only consider the relationships within the class withoutenough attention to the distance between classes. When it comes to fuzzy boundary data, misclassification problems inboundary may easily occur due to the overlapping of the centers. To solve the above problems, this paper introduces amaximum penalty term between classes in high-dimensional feature space and the control parameter λ based on the KPCM.The new proposed algorithm which constructs a new objective function is called the Maximum center interval Kernel PossibilisticC-Means(MKPCM)clustering algorithm. The algorithm makes the distance between the centers maximum bythe maximum penalty term between centers and through the control parameter λ , it effectively avoids the event of tooclose centers or even overlaps. Numerical experimental results demonstrate its favorable performance especially in thematter with fuzzy boundary. In addition, it shows distinct advantages in the application of image segmentation comparedto the traditional cluster methods.