融入局部信息的直觉模糊核聚类图像分割算法

An Intuitionistic Kernel-based Fuzzy C-means Clustering Algorithm with Local Information for Image Segmentation

针对传统直觉模糊C均值聚类(Intuitionistic Fuzzy C-means,IFCM)的图像分割算法对噪声和初始聚类中心敏感,导致聚类精度不高和迭代次数多的问题,提出一种结合局部信息的直觉模糊核聚类的图像分割算法。在该算法中,首先采用基于直方图的方法确定聚类中心初始值,解决算法对聚类中心的初始值敏感的问题;其次,利用核函数将待分类数据集映射到高维非线性空间,改善数据的线性可分性,同时在目标函数中引入局部灰度信息和局部空间信息,提高直觉模糊聚类的分类精度。实验结果表明,提出算法能减少迭代次数,提高聚类精度,能有效对图像进行分割;无论在对图像分割还是在聚类有效性上,提出算法都要优于传统的模糊聚类算法,如模糊C均值聚类(Fuzzy C-means,FCM)、模糊核均值聚类(Kernel-based fuzzy c-means,KFCM)、引入空间信息的直觉模糊C均值聚类(Intuitionistic Fuzzy C-means with spatial constraints,IFCM-S)、模糊空间聚类(Fuzzy Local Information C-means,FLICM)、直觉模糊C均值聚类(Intuitionistic Kernel-based Fuzzy C-means,IFKCM)等。

Intuitionistic fuzzy c-means（ IFCM） clustering segmentation algorithm is sensitive to noise and initial clustering center,which leads to the problem of low accuracy and huge iterations. A novel intuitionistic kernel-based fuzzy C-means clustering algorithm which takes into account local information is proposed for image segmentation. The new histogram-based algorithm is used for initial clustering,which resolve the problem that centroid of cluster is sensitive to the initial values.Furthermore,it incorporates the advantage of intuitionistic fuzzy sets theory and kernel function,and kernel function is adopted to map intuitionistic fuzzy data samples to a high-dimension feature space,which is helpful for clustering. At the same time,through incorporating the local gray information and local spatial information in the objective function,an intuitionistic kernel-based fuzzy c-means clustering algorithm with local information is obtained by mathematical deduction. The experiments demonstrate that the proposed algorithm can reduce the number of iterations,improve the classification accuracy,and segment the image effectively. Both in image segmentation and the effectiveness of clustering,the performance of the proposed algorithm is superior to conventional fuzzy clustering methods,including fuzzy c-means（ FCM）,kernel-based fuzzy c-means（ KFCM）,intuitionistic fuzzy c-means with spatial constraints（ IFCM-S）,fuzzy local information c-means（ FLICM） and intuitionistic kernel-based fuzzy c-means（ IKFCM） algorithms.