隐马尔科夫模糊C均值聚类分割算法改进  

An improved hidden Markov fuzzy C-means clustering segmentation algorithm

作　　者：吴成茂[1] 何晶[1] 王辉[2] 

机构地区：[1]西安邮电大学电子工程学院,陕西西安710121 [2]西安邮电大学自动化学院,陕西西安710121

出　　处：《西安邮电大学学报》2017年第1期44-49,共6页Journal of Xi’an University of Posts and Telecommunications

基　　金：国家自然科学基金重点项目(61136002);陕西省自然科学基金资助项目(2014JM8331;2014JQ5183;2014JM8307);陕西省教育厅科学研究计划资助项目(2015JK1654)

摘　　要：对隐马尔科夫模糊C均值算法加以改进,以提高其对加噪图像的分割质量。在初始化之前,先对图像进行中值滤波,以降低均值漂移算法产生的过分割现象,再对原有算法迭代所得隶属度进行均值滤波,以增强图像分割的抗干扰能力。对标准灰度图像添加噪声,验证改进算法的性能。实验结果显示,经过改进,算法的分割结果更稳定,边界更平滑,且对噪声具有较强鲁棒性。The hidden Markov fuzzy C-means algorithm is amended to improve its performance in image segmentation.Before initialization,the image is median-filtered in order to reduce the over segmentation caused by the mean shift algorithm.moreover,the membership degree iteratively obtained from the original algorithm is mean-filtered in order to enhance the anti interference ability of the image segmentation.The performance of the improved algorithm is verified by adding noise to the standard gray scale images,which showes that,after improvement,the segmentation results are more stable,the boundary is smoother,and the algorithm is robust against the noise.

关 键 词：隐马尔科夫模糊C均值算法 均值漂移 信噪比 中值滤波 均值滤波 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]