摘要
加权模糊C均值(WFCM)算法是在模糊C均值(fuzzyC-Means,FCM)算法的基础上提出的,它为不同的样本添加了不同的权值,从而改善了聚类效果。然而传统的加权模糊C均值算法具有对噪声非常敏感的缺陷,于是本文提出了一种结合Gibbs随机场的改进的WFCM算法(G-WFCM)。根据Gibbs随机场概率分布构造了一个Gibbs空间约束场,通过用Gibbs空间约束场为WFCM施加空间约束的方法来减小噪声对分割结果的影响。文中给出的人脑MRI图像分割实验证明,本文提出的G-WFCM算法具有比原WFCM算法更强的抗椒盐噪声能力。
Weighed Fuzzy C-Means (WFCM) algorithm is based on Fuzzy C-Means (FCM) algorithm, which adds different weighs to different samples hence improved the clustering results. But traditional WFCM algorithm is very sensitive to noise, so we propose a new WFCM algorithm combined with Gibbs random field (GRF). We construct a Gibbs spatial constraint field based on GRF and Gibbs distribution, and bring spatial constraint to bear on WFCM, which minished the negative effect for segmentation caused by noise. The illustrating example of MRI image segmentation in this paper indicates this algorithm is more resistible than WFCM under "salt & pepper" noise.
出处
《电子测量技术》
2007年第11期190-192,共3页
Electronic Measurement Technology
基金
国家博士点基金资助项目(20040699015)
西北工业大学研究生创业种子基金资助项目(Z200631).