结合局部和非局部信息的FCM图像分割算法  

An FCM Algorithm Incorporating with Local and Non-local Information for Image Segmentation

作　　者：宋小鹏[1] 李国熊 张权[1] 桂志国[1] SONG Xiao-peng;LI Guo-xiong;ZHANG Quan;GUI Zhi-guo(Shanxi Key Laboratory of Biomedical Imaging and Big Data,North University of China,Taiyuan 030051,China)

机构地区：[1]中北大学生物医学成像与影像大数据山西省重点实验室,太原030051

出　　处：《火力与指挥控制》2020年第10期28-33,共6页Fire Control & Command Control

摘　　要：针对结合局部信息模糊C均值聚类算法(FLICM)模糊因子只含局部信息且权重系数仅由欧式距离决定,对噪声图像的边缘细节分割不准确的问题,提出了一种结合图像局部和非局部信息的改进FLICM算法。通过融入图像非局部信息,模糊因子由局部和非局部信息共同决定,两者的权重系数能根据像素结构信息的不同而自适应地变化。实验表明,改进算法在抑制噪声的同时,能更多地对图像边缘细节进行保护,有效地提高了对噪声图像的分割准确率,并有着更好的模糊划分效果。As the fuzzy factor only includes local information and its weighting coefficient is just based on Euclidean distance,the edges,and details in noise image can't be segmented accurately by Fuzzy Local Information C-means Clustering Algorithm(FLICM),an improved FLICM algorithm incorporating both local and non-local information is presented.By introducing non-local information of image,the fuzzy factor is redesigned.The fuzzy factor of the improved algorithm is determined by both local and non-local information,and their weighting coefficients vary adaptively according to pixels structure information.Experiment results indicate that the improved algorithm can preserve more edges and details while the noises are restrained,and the segmentation accuracy of noise image is improved effectively with better fuzzy partition result.

关 键 词：模糊C均值聚类 图像分割 非局部 模糊因子 边缘细节 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置] E919[自动化与计算机技术—控制科学与工程]