自适应细菌觅食的FCM聚类优化算法研究

刘志鹏 胡亚琦 张卫卫

摘  要： 在初始化FCM聚类算法时，聚类类别数需要手动去设置，并随机初始聚类中心，导致此算法极其容易陷入局部最优值。通过利用改进的细菌觅食算法，进行FCM算法的聚类中心的初始化，解决FCM算法对初始聚类中心敏感的问题;通过一些有效性的指标，对FCM算法和优化FCM算法进行评估，指标说明了优化FCM算法更好。在仿真实验中，将优化FCM算法和标准FCM算法用到多类图像分割中，进行了图像分割的准确性和实时性的比较，且验证了所述的优化算法的实时性。

关键词： FCM算法; 自适应细菌觅食; 聚类优化; 算法评估; 仿真实验; 图像分割

中图分类号： TN911.73?34; TP391.4             文献标识码： A                      文章編号： 1004?373X（2020）06?0144?05

Research on FCM clustering optimization algorithm for self?adaptive bacterial foraging

LIU Zhipeng， HU Yaqi， ZHANG Weiwei

（School of Electronic and Information Engineering， Lanzhou Jiaotong University， Lanzhou 730070， China）

Abstract： As the initialization of the FCM clustering algorithm is performed， the number of clustering categories needs to be set manually and the clustering center is initialized randomly， which makes this algorithm extremely easy to fall into the local optimum. The clustering center of FCM algorithm is initialized by means of the improved bacterial foraging algorithm to solve the problem that FCM algorithm is sensitive to the initial clustering center. The FCM algorithm and the optimized FCM algorithm are evaluated with some validity indexes， which shows that the optimized FCM algorithm is better. In the simulation experiments， the optimized FCM algorithm and the standard FCM algorithm were used in the multi?class image segmentation to compare their accuracy and real?time performance for image segmentation， by which the real?time performance of the proposed optimization algorithm was verified.

Keywords： FCM algorithm; self?adaptive bacterial foraging; clustering optimization; algorithm assessment; simulation experiment; image segmentation