二维变分模态分解联合快速非局部均值的医学超声图像去噪方法  

MEDICAL ULTRASOUND IMAGE DENOISING ALGORITM BASED ON TWO-DIMENSIONAL VARIATIONAL MODAL DECOMPOSITION AND FAST NON-LOCAL MEAN

作　　者：闫洪波[1] 刘霈 徐洋[2] 杨文英 赵蓬勃 Yan Hongbo;Liu Pei;Xu Yang;Yang Wenying;Zhao Pengbo(School of Mechanical Engineering,Inner Mongolia University of Science and Technology,Baotou 014010,Inner Mongolia,China;The First Affiliated Hospital of Baotou Medical College,Baotou 014010,Inner Mongolia,China;Sinopharm North Hospital(The Third Affiliated Hospital of Baotou),Baotou 014030,Inner Mongolia,China)

机构地区：[1]内蒙古科技大学机械工程学院,内蒙古包头014010 [2]包头医学院第一附属医院,内蒙古包头014010 [3]国药北方医院(包医三附院),内蒙古包头014030

出　　处：《计算机应用与软件》2023年第3期260-265,共6页Computer Applications and Software

基　　金：科技部创新方法工作专项项目(2017M010660-03);内蒙古自治区自然科学基金项目(2020LH05023)。

摘　　要：针对医学超声图像成像过程中出现的斑点噪声,提出一种基于二维变分模态分解(2D-VMD)联合快速非局部均值(FNLM)的去噪算法。该算法利用2D-VMD获得一系列高频至低频的本征模态分量;对保留绝大部分信息的低频模态进FNLM滤波,从而实现对超声图像的去噪。实验结果与传统滤波方法进行对比,提出算法在去噪的同时较好地保留边缘信息,有更好的客观评价参数,尤其是在高噪声方差中去噪效果明显优于其他算法。Aimed at speckle noise in medical ultrasound image imaging,a denoising algorithm based on two-dimensional variational modal decomposition(2D-VMD)combined with fast non-local mean(FNLM)is proposed.The algorithm used 2D-VMD to obtain a series of high-frequency to low-frequency eigenmode components.The FNLM filtered the low-frequency modes that retained most of the information,so as to achieve denoising of ultrasound images.The experimental results were compared with traditional filtering methods.The proposed algorithm can better retain the edge information while denoising,and has better objective evaluation parameters,especially in the high noise variance,the denoising effect is obviously better than other algorithms.

关 键 词：斑点噪声 二维变分模态分解 快速非局部均值 客观评价参数 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]