一种改进的AFCM聚类算法

An improvement on AFCM clustering algorithms

模糊C-均值(FCM)聚类算法在图象处理和模式识别领域中得到了广泛的应用,但由于FCM算法在大数据集的情况下需要消耗大量的CPU时间而使用户感到十分不便。本文对近似的模糊C-均值(AFCM)算法进行了改进,提出了一种改进的AFCM(IAFCM)聚类算法。对一个128×128的彩色数字图象进行FCM、AFCM、IAFCM算法聚类,结果表明,IAFCM算法所用的时间约为AFCM算法的二分之一,仅为FCM算法的十三分之一。

The fuzzy c-means (FCM) clustering algorithms are widely appplied in imageprocessin and pattern recognition. However, for large data sets the FCM algorithm, calledIAFCM, which can improve the approximate fuzzy c-means (AFCM) algorithm. The ex-perimental results show that the net effect of IAFCM implementation is that the CPU timeis reduced to about a half of the time required for AFCM implementation, or to about onethirteenth of the time required for FCM implementation.