基于改进模糊C均值聚类与Otsu的图像分割方法  

Image Segmentation Method Based on Improved Fuzzy C-means Clustering and Otsu Maximum Variance

作　　者：王勋[1] 李廷会[1] 潘骁 田宇 WANG Xun;LI Tinghui;PAN Xiao;TIAN Yu(College of Electronic Engineering,Guangxi Normal University,Guilin Guangxi 541004,China;Department of Automotive and Information Engineering,Guangxi ECO-engineering Vocational and Technical College,Liuzhou Guangxi 545004,China)

机构地区：[1]广西师范大学电子工程学院,广西桂林541004 [2]广西生态工程职业技术学院汽车与信息工程系,广西柳州545004

出　　处：《广西师范大学学报（自然科学版）》2019年第4期68-73,共6页Journal of Guangxi Normal University:Natural Science Edition

基　　金：国家自然科学基金(21327007);广西师范大学青年基金(2017QN002);高新企业技术发展(20170113-1)

摘　　要：针对背景与前景颜色差别较小的原木图像分割效果不理想的情况,本文给出了模糊C均值聚类与Otsu相结合的图像分割方法。该方法首先以标准原木数据库为样本,之后使用模糊C均值聚类算法把背景与前景颜色差别较小的原木样本图像分割成2类,其次利用准则函数找出前景分割结果,最后把该结果作为Otsu算法的输入,对原木样本图像进行再次分割。实验结果表明,本文研究的算法比单独使用模糊C均值聚类算法、Otsu和同类算法有较好的分割效果和较高的分割准确率,边缘信息保留较好,平均分割准确率提高2个百分点。To solve the problem that the segmentation effect of log image with small difference between background and foreground color is unsatisfactory,an image segmentation method based on fuzzy C-means clustering and Otsu is presented.Firstly,standard log database is used as sample and the log sample image with small difference between background and foreground color is segmented into two categories by using the fuzzy C-means clustering algorithm.Secondly,the criterion function is used to obtain the result of foreground segmentation.Finally,the result is used as input of Otsu algorithm to form the final segmentation of log sample image.The experimental results show that the proposed algorithm is better than the single fuzzy C-means clustering algorithm.Otsu and similar algorithms have better segmentation effect and higher segmentation accuracy,the edge information is well preserved and the average segmentation accuracy is increased by 2 percentage points.

关 键 词：图像分割 模糊C均值聚类 OTSU 准则函数 分割准确率 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]