基于改进的模糊c均值聚类水凝胶图像分割  

The segmentation of hydrogel image based on improved Fuzzy c-means clustering

作　　者：李艳[1] 杜宇人[1] 沈鑫[1] 

机构地区：[1]扬州大学信息工程学院,江苏扬州225127

出　　处：《扬州大学学报（自然科学版）》2012年第1期60-64,共5页Journal of Yangzhou University：Natural Science Edition

基　　金：国家自然科学基金资助项目(20299030)

摘　　要：基于实验图像中目标与背景区域具有明显的色差和形状差异且其直方图有明显双峰的特性,提出一种改进的模糊c均值聚类和数学形态学相结合的算法,将目标图像有效地分割出来.首先依据像素邻域间的约束属性以及二维直方图对角线元素含有稳定图像信息的特点,在模糊c均值聚类算法基础上对聚类目标函数进行改进,并将二维直方图对角函数引入隶属度函数中,最终将目标区域有效地分割出来.实验结果表明:该方法具有识别精度高、计算量小、去噪能力强的特点.The histogram of image in the paper has two obvious peaks the image contains different characteristics which exist in gray and edge between target and background regions.An improved algorithm combining improved FCM and morphological is proposed.Based on the properties of the constraints between pixel neighborhood and two-dimensional histogram with stable image information,the cluster objective function is improved based on fuzzy c-means clustering algorithm,and two-dimensional histogram angle function is introduced into membership function.Then the segmentation image treated with the improved fuzzy c-means is carried on mathematical morphology.Finally the target region is segmented accurately.Experimental results show that the improved method has the properties of high recognition accuracy,less computation,strong de-noising ability.

关 键 词：水凝胶 图像分割 FCM算法 MFCM算法 数学形态学 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]