改进TLBO优化人工鱼群算法实现FCM图像分割

Artificial fish swarm algorithm with improved TLBO algorithm to achieve FCM image segmentation

FCM算法是一种比较流行的模糊聚类算法,将其运用到图像分割领域中,可以更加客观地对像素进行分类,达到较好的分割效果。但是传统的FCM算法存在对初始聚类中心较为依赖的问题,一般情况下,初始聚类中心是随机选取的,如果选择的不合适,则很容易导致算法得不到全局最优解。针对此问题,提出了改进方法,对人工鱼群算法(AFSA)进行改进,利用人工鱼群算法良好的全局搜索功能,同时融合改进的教学优化算法(TLBO),用于搜索FCM算法的初始聚类中心。通过实验证明,提出的算法不仅有效地改善了人工鱼群算法后期寻优速度慢、不精确的问题,并且与标准FCM算法和人工鱼群算法优化FCM算法相比,在迭代次数和运行时间方面,平均下降了14%~55%,分割正确率平均提高了7%~10%,从而在图像分割方面具有更快的分割速度和更优的分割效果。

FCM algorithm is a popular fuzzy clustering algorithm,which can be applied to image segmentation field to classify pixels more objectively and achieve better segmentation effect.However,the traditional FCM algorithm is relatively dependent on the initial clustering center.In general,the initial clustering center is randomly selected.If the selection is not appropriate,it is easy for the algorithm to fail to obtain the global optimal solution.To solve this problem,this paper proposes the following improvement methods:The Artificial Fish Swarm Algorithm(AFSA)is improved to search the initial clustering center of FCM algorithm by combining the improved Teaching⁃Learning⁃Based Optimization(TLBO)algorithm with the good global search function of artificial fish swarm algorithm.Experiments show that the algorithm proposed in this paper not only effectively improves the problem of slow optimization speed and inaccurate optimization results of artificial fish swarm algorithm in the later stage,but also compared with the standard FCM algorithm and the artificial fish swarm optimized FCM algorithm,in terms of iteration times and running time,the average number of iterations decreased by 14%~55%,and the average segmentation accuracy increased by 7%~10%,thus achieving faster segmentation speed and better segmentation effect in image segmentation.