计算机工程与应用 ›› 2008, Vol. 44 ›› Issue (12): 186-188.

• 图形图像、模式识别 • 上一篇    下一篇

基于小波包和改进的FCM的医学图像分割

吕 回,李 峰,徐 琼   

  1. 长沙理工大学 计算机与通信工程学院,长沙 410076
  • 收稿日期:2007-08-07 修回日期:2007-11-12 出版日期:2008-04-21 发布日期:2008-04-21
  • 通讯作者: 吕 回

Medical image segmentation based on wavelet packet and improved FCM

LV Hui,LI Feng,XU Qiong   

  1. College of Computer and Communication Engineering,Changsha University of Science and Technology,Changsha 410076,China
  • Received:2007-08-07 Revised:2007-11-12 Online:2008-04-21 Published:2008-04-21
  • Contact: LV Hui

摘要: 提出应用最优小波包变换对磁共振颅脑图像做分解,以各子带小波包系数的能量形成纹理特征集;并运用基于核函数的模糊C均值聚类算法(Kernel-Based Fuzzy C-means Algorithm,KFCM)对所提取到的特征集进行聚类分析,从而实现了对磁共振颅脑图像的有效分割。实验证明应用KFCM算法做分割的收敛速度和抗噪性明显优于FCM算法。

关键词: 磁共振颅脑图像, 医学图像分割, 最优小波包变换, 纹理特征, KFCM算法

Abstract: A method is proposed for segmenting MR brain images using the best wavelet packet transform,and the eigenvectors are come into using the energy of wavelet packets’ coefficients.Then the Kernel-Based Fuzzy C-means Algorithm (KFCM) is used for clustering analysis of the eigenvectors. Finally,the MR brain image is segmented effectively by our method.The experiment proves that KFCM for segmentation is more robust in convergence and anti-noise than FCM.

Key words: magnetic resonance imaging of brain, medical image segmentation, the best wavelet packet transform, texture feature, KFCM algorithm