考虑小波系数相关性的超声图像降噪算法  

Algorithm considering correlation of wavelet coefficients for ultrasound image denoising

作　　者：宋坤坡[1] 夏顺仁[1] 徐清[2] 

机构地区：[1]浙江大学生物医学工程教育部重点实验室,浙江杭州310027 [2]浙江大学校医院,浙江杭州310027

出　　处：《浙江大学学报（工学版）》2010年第11期2203-2208,共6页Journal of Zhejiang University：Engineering Science

基　　金：国家安全重大基础研究资助项目(5132103ZZT14B);国家自然科学基金资助项目(60772092)

摘　　要：为了抑制超声图像中的斑点噪声,提出一种考虑小波系数尺度间相关性的超声图像降噪算法.该算法采用Rayleigh分布对超声图像斑点噪声的统计特性建模和Laplacian分布对小波系数的统计特性进行建模,进而利用贝叶斯最大后验的方法获得对无噪图像的估计.为了更好地保留图像细节,在阈值计算过程中,该算法通过考虑下一尺度对应的小波系数来构造一个尺度间相关因子.实验结果表明,所提出算法在有效减少斑点噪声的同时,更好地保持了图像边缘和细节.In order to suppress the speckle noise in ultrasound image,an algorithm considering the inter-scale correlation of wavelet coefficients was proposed.The algorithm used the Rayleign distribution to model the statistics of the speckle noise,and the Laplacian distribution was used to model the statistics of the wavelet coefficients of ultrasound image,and the estimation of the noise-free image through Bayesian maximum a posteriori was finally obtained.In order to preserve the details better,an inter-scale correlation factor was adopted in the calculation of the threshold,which was constructed by considering corresponding wavelet coefficients in the next scale.Experiments show that the proposed algorithm suppresses speckle noise well,while retaining the edges and details much better.

关 键 词：超声图像 斑点噪声抑止 贝叶斯最大后验估计 小波系数相关性 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]