基于混沌粒子群和模糊聚类的图像分割算法

摘要

模糊C-均值聚类算法(FCM)是一种结合模糊集合概念和无监督聚类的图像分割技术,适合灰度图像中存在着模糊和不确定的特点；但该算法受初始聚类中心和隶属度矩阵的影响,易陷入局部极小.利用混沌非线性动力学具有遍历性、随机性等特点,结合粒子群的寻优特性,提出了一种基于混沌粒子群模糊C-均值聚类(CPSO-FCM)的图像分割算法.实验证明,该方法不仅具有防止粒子因停顿而收敛到局部极值的能力,而且具有更快的收敛速度和更高的分割精度.%Fuzzy C-means(FCM) clustering algorithm was an effective image segmentation algorithm which combined the con-cept of fuzzy sets and unsupervised clustering. And it suited for the uncertain and ambiguous characteristic in intensity image. But it was sensitive to initial clustering center and membership matrix and likely converged into the local minimum, which caused the quality of image segmentation lower. By using of the properties-ergodicity, randomicity of chaos, this paper pro-posed a new image segmentation algorithm, which combined the chaos particle swarm optimization (CPSO) and FCM cluste-ring. Experimental results prove this method not only has the ability to prevent the particles to convergence to local optimum because of standstill, but also has faster convergence and higher accuracy of segmentation.