一种具有边缘保持特性的超声图像小波域阈值去噪新方法

Novel speckle reduction for medical ultrasound images based on edge preservation

超声图像去噪是医学图像处理的研究热点之一 ,基于小波域阈值去噪技术及阈值选取方法的分析 ,提出一种新的医学超声图像小波域阈值去噪方法。这种方法采用半_软阈值去噪技术和广义交叉确认函数寻找阈值 ,在有效去噪的同时较好地保留了图像边缘细节。首先 ,把对数超声图像小波分解 ;然后 ,基于广义交叉确认函数寻找最小均方误差意义上的近似最优阈值 ,对所有的高频段采用半_软阈值去噪 ;最后 ,经小波反变换和指数变换获得去噪后的超声图像 ,文末对超声图像小波域阈值去噪方法作出定性比较 ,并对算法的去噪性能给出定量分析。仿真实验和实际测试结果表明此方法是有效的、可行的。

A novel speckle reduction for medical ultrasound images is presented based on the study of wavelet domain threshold-denoising and threshold-choosing. First, the logarithmic transform of the original image is decomposed into the multiscale wavelet domain. Then, semi-soft thresholding is used to reduce speckle noise. Finally, the denoised image is achieved by the invert DWT and the exponient transform of the estimated wavelet coefficients. Current state-of-the-art soft and hard thresholding methods based on universal threshold have been applied in actual ultrasound medical images. Compared with our method, the achieved performance improvement is quantified. Performance of the proposed method has been tested on Ultrasound images. The results show the method effectively reduces the speckle while preserving the edges of the original image.