显著信息引导的直觉空间模糊聚类图像分割

Intuitionistic spatial fuzzy clustering for image segmentation guided by significant information

为克服直觉模糊C-均值(IFCM)聚类算法应用于图像分割时,易受噪声影响,且对聚类中心初始值敏感的缺陷,给出显著信息引导的直觉空间模糊聚类图像分割方法。使用图像的显著信息初始化聚类中心,能够很大程度地防止算法陷入局部最优;将改进的融合局部空间信息的模糊因子引入到IFCM聚类算法中,可提升算法的抗噪性能。实验结果表明所给方法能在多种含噪声图像上得到较好的分割效果。

In order to overcome the defects of intuitionistic fuzzy C-means （IFCM）clustering algorithm for image segmentation, which is easily affected by noise and sensitive to the initial clustering centers, an intuitionistic spatial fuzzy clustering image segmentation algorithm guided by significant information is presented. Using the significant information of the image to initialize the clustering center can largely prevent the algorithm from failing into local optimum. An improved fuzzy factor that combines the local spatial information is introduced into the IECM clustering algorithm to improve the algorithm＇s robustness against noise. Experimental results show that the proposed method can get better segmentation results on many kinds of noisy images.