融合结构特征的增强型FCM图像分割算法  

Enhanced FCM Algorithm Combined with Structure Feature for Image Segmentation

作　　者：崔兆华[1] 张萍[2] 李洪军[3] 高立群[1] 

机构地区：[1]东北大学信息科学与工程学院,辽宁沈阳110819 [2]鞍山师范学院,辽宁鞍山114005 [3]白城医学高等专科学校,吉林白城137000

出　　处：《东北大学学报（自然科学版）》2013年第7期922-926,共5页Journal of Northeastern University(Natural Science)

基　　金：国家自然科学基金资助项目(51005042);国家高技术研究发展计划项目(2012AA062002);中央高校基本科研业务费专项资金资助项目(N100403005)

摘　　要：为了使基于模糊C均值(FCM)聚类的图像分割算法对复杂图像更具适用性,将图像结构特征融合到增强型FCM算法.首先,对原始图像进行均值滤波,将滤波结果与原始图像进行线性叠加形成新的输入图像.其次,采用二维Gabor滤波函数提取新的输入图像的纹理结构特征,以此代替灰度特征来衡量节点间的相似性.最后,采用一种改进的节点间距离度量公式来计算图像中节点与聚类中心点的差异.仿真结果表明,对结构复杂的图像所提算法获得了更加精确的分割结果.To improve the ability of fuzzy C-means clustering algorithm（FCM） for complex texture structure images,a new fuzzy C-means clustering algorithm（En-FCM） was proposed by combining image structure features.Firstly,the input image was mean-filtered,and the filtered image was added to the original image to form the new image for the subsequent operations.Secondly,the 2-D Gabor filtering function was adopted to extract texture structure feature for the new images to replace the gray level similarity measurement in the traditional FCM algorithm.Finally,a new distance measure function was proposed to calculate the distance betw een the nodes and the clusters.The simulation results show ed that more precise segmentation results could be obtained from complicated texture structure images using the presented algorithm.

关 键 词：图像分割 模糊C均值聚类 均值滤波 纹理特征 二维Gabor滤波器 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]