基于分层的直觉模糊C均值聚类图像分割算法  

Intuitionistic fuzzy C-means clustering algorithm based on hierarchy for image segmentation

作　　者：池桂英 王忠华[1,2] 

机构地区：[1]南昌航空大学信息工程学院,江西南昌330063 [2]南昌航空大学无损检测技术教育部重点实验室,江西南昌330063

出　　处：《计算机工程与设计》2017年第12期3368-3373,共6页Computer Engineering and Design

基　　金：国家自然科学基金项目(61362036);2015年研究生创新专项基金项目(YC2015-S320);江西省自然科学基金项目(20132BAB201024);无损检测教育部重点实验室开放基金项目(ZD201429005)

摘　　要：传统模糊C均值聚类算法及其相关改进算法应用于图像分割时,未充分利用像素点的邻域信息,导致图像的分割结果不理想、运行效率偏低等问题,为此,提出基于分层的直觉模糊C均值聚类图像分割算法。采用分层技术将图像划分为多个不同区域,计算其相应的聚类中心;构造融合相邻像素的相关性与直觉模糊集理论的目标函数,求解直觉模糊隶属度矩阵;根据像素的隶属度修正图像分层,直到符合聚类准则。实验结果表明,该算法取得了良好的图像分割效果,提升了图像分割效率。When the traditional fuzzy C-means clustering algorithm and its improved algorithm were applied to image segmentation,due to not making full use of the neighborhood information,the image segmentation result is not ideal and the running efficiency is low.To overcome these drawbacks,an intuitionistic fuzzy C-means clustering algorithm based on hierarchy for image segmentation was presented.The image was divided into several different regions using the hierarchical technique.The cluster center was computed in the corresponding region.The objective function was constructed with the theory of intuitionistic fuzzy sets and the relevance between adjacent pixels to acquire the intuitionistic fuzzy membership matrix.According to the membership matrix,the image layer was modified continuously until it satisfied the clustering criterion.Experimental results show that the algorithm can not only get better the image segmentation result,but also significantly improve the efficiency of image segmentation.

关 键 词：图像分割 模糊C均值 直觉模糊集 分层技术 缺陷图像 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]