基于双密度双树小波变换的超声图像降噪  

Denoising of Ultrasonic Image via Double Density Dual Tree Wavelet Combined with Bivariate Shrinkage with Local Variance Estimation

作　　者：李鹏[1] 喻罡[1] 冀晓燕[1] 卞正中[1] 

机构地区：[1]西安交通大学生命科学与技术学院生物医学工程系,西安710049

出　　处：《系统仿真学报》2007年第24期5797-5801,共5页Journal of System Simulation

基　　金：国家自然科学基金(60271022;60571006)

摘　　要：针对去除斑点噪声提高超声图像质量的问题,提出双密度双树离散小波变换(DD-DT DWT)结合局部方差估计的双变量收缩阈值函数(BFS)的图像降噪改进算法实现超声图像降噪。首先将原始图像用DD-DT DWT进行多尺度分解,根据噪声模型和小波子父代系数确定的局部边缘方差估计阈值,利用子父代小波系数相关性构成的双变量阈值函数,对图像16个方向的小波系数进行非线性自适应的处理,最后重建降噪后的图像。用仿真和真实数据对此算法进行验证,并与其他小波降噪系统的性能比较,结果分析表明噪声图像经该算法降噪后,图像性能指标均有提高,不仅有效的实现图像降噪,而且较好的保留图像细节。The improved algorithm was proposed for reducing the ultrasonic image speckle noise and ameliorating image quality, in which the double density dual tree discrete wavelet transform （DD-DT DWT） was combined with the bivariate shrinkage function （BSF） with local variance estimation, The original image was firstly decomposed by the DD-DT DWT, then the threshold was estimated according to the noise model and the marginal variance of the local noisy wavelet coefficients and their parent coefficients, and the wavelet coefficients were shrunk by the BSF related to dependence of parent and son wavelet coefficients, in which all of 16 orientations were nonlinear processed adaptively, finally the denoised image was reconstructed by all the update coefficients. The improved algorithm was tested by simulated and actual data, and was compared with other wavelet denoising algorithms. The results indicate that the performances of the denoised image via the proposed algorithm are coincidently improved, and that the effective denoising of image and the preserving of particular are simultaneously obtained.

关 键 词：双密度双树小波 双变量阈值 局部方差估计 系数相关性 超声图像降噪 斑点噪声 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]