基于EM算法参数估计的各向异性扩散超声图像的去噪

被引:5
作者
余锦华 [1 ]
汪源源 [1 ]
施心陵 [2 ]
机构
[1] 复旦大学电子工程系
[2] 云南大学电子工程系
关键词
混合分布模型; EM算法; 各向异性扩散; 超声图像; 去噪;
D O I
10.16289/j.cnki.1002-0837.2007.03.010
中图分类号
TN911.73 [图像信号处理];
学科分类号
0711 ; 080401 ; 080402 ;
摘要
目的提出一种各向异性扩散滤波器的扩散参数选取方法,提高滤波器的灵活性和稳定性。方法使用二状态的瑞利、高斯混合分布对超声图像灰度分布进行拟合,并采用期望值最大化(expectation maximization,EM)算法实现混合分布的分解;根据分解结果预测图像中斑点噪声均匀分布的区域;通过对均匀区域统计特性的分析获取各向异性扩散的扩散参数。结果通过与两种改进扩散参数选取的滤波方法对比,基于EM算法的混合分布分解能够准确地估计扩散参数,使滤波结果在噪声消除和边缘保持上达到有效的平衡。结论基于EM算法参数估计的各向异性扩散是一种有效的超声图像去噪方法。
引用
收藏
页码:198 / 204
页数:7
相关论文
共 10 条
[1]  
Calculation of pressure fields from arbitrarily shaped,apodized,and excited ultra-sound transducers. Jensen JA,Svendsen NB. IEEE Trans Ultrason,Ferro-elec,Freq Contr . 1992
[2]  
Digital image processing. Gonzalez RC,Woods RE. . 2003
[3]  
De-noising by soft-thresholding. Donoho DL. IEEE Transactions on Information Theory . 1995
[4]  
An adaptive speckle suppression filter for medical ultrasonic imaging. Karaman M,Kutay MA,Bozdagi G. IEEE Transactions on Medical Imaging . 1995
[5]  
Some fundamental properties of speck-le. Goodman J W. Journal of the Optical Society of America . 1978
[6]  
Speckle reducing anisotropic diffusion. YU Yongjian,Acton ST. IEEE Transactions on Image Processing . 2002
[7]  
Nonlinear multi-scale wavelet diffusion for speckle suppression and edge enhancement in ultrasound images. Yue Yong,Croitoru MM,Bidani A,et al. IEEE Transactions on Medical Imaging . 2006
[8]  
Edge detection in ultrasound imagery using the instantaneous coefficient of variation. YU Yongjian,Acton ST. IEEE Transactions on Image Processing . 2004
[9]  
Scale space and edge detection using anisotropic diffusion. Perona P,Malik J. IEEE Trans Pattern and Ma-chine Intell . 1990
[10]  
Speckle in ultrasound B-mode scans. Burckhardt CB. IEEE Transactions on Sonics and Ultrasonics . 1978