基于优化初始聚类中心的改进WFCM图像分割算法

摘要

针对模糊C均值聚类算法在分割图像时容易陷入局部最优的问题,提出一种改进的加权模糊C均值聚类图像分割算法.该算法借助图像直方图势函数,找出图像直方图上潜在的图像分割点;再基于 Fisher 判别思想构造关于模糊分割初始聚类中心的最优化问题,以期求取合理初始聚类中心,避免算法陷入局部最优.仿真结果表明,改进的加权模糊聚类图像分割算法能够有效避免陷入局部最优,具有更高的分割效率.%In order to effectively avoid weight fuzzy C-means (WFCM) segment algorithm convergence to local optimum, an improved WFCM image segmentation algorithm based on optimization of the initial clustering center is proposed in this paper. With aid of image histogram potential function, potential segmentation points are found; and based on the Fisher discrimination; the optimization problem of fuzzy C-means segmentation algorithm is constructed. Next, the optimization problem is solved for obtain reasonable initial clustering centers. Then, WFCM is using the centers to finish image segmentation. The experiments results show that the improved algorithm for image segmentation based WFCM more effective convergence to global optimal solution segmentation than standard WFCM in image segmentation.