改进的模糊C-均值算法在图像分割中的应用  

Application of Improved Fuzzy C-Means Algorithm in Image Segmentation

作　　者：周丹[1] 肖满生[1] 刘丽红[1] 姚慧丹 

机构地区：[1]湖南工业大学计算机与通信学院,湖南株洲412007

出　　处：《湖南工业大学学报》2014年第5期79-83,共5页Journal of Hunan University of Technology

基　　金：国家自然科学基金资助项目(61170102);湖南省教育厅重点基金资助项目(12A042)

摘　　要：针对传统的模糊C-均值算法在图像分割中存在的缺陷,提出了一种基于点密度函数加权的模糊C-均值聚类算法。将图像像素的点密度函数作为权值,并依据类间相关度定义了一个聚类有效性函数用以确定最佳聚类数,结合聚类有效性完成对图像的分割。理论分析和对比试验表明,该算法在一定程度上克服了模糊均值算法的缺陷,在图像分割中具有良好的分类精度。For the defects of traditional fuzzy C-means algorithm in image segmentation, a weighted fuzzy C-means clustering algorithm based on dot density function are proposed. Takes the dot density function of image pixels as the weight, and on the basis of inter-class correlation defines a cluster validity function to determine the optimal number of clusters and combines with cluster validity to complete the effective image segmentation. Theoretical analysis and com- parative experiments show that the algorithm overcomes the shortcomings of fuzzy means algorithm to some extent and has good classification accuracy in image segmentation.

关 键 词：模糊C-均值 点密度函数 聚类有效性 图像分割 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]