保留细节的迭代非局部均值超声波图像去噪  

Detail preserving iterative non-local means algorithm for de-speckling of ultrasound images

作　　者：刘祝华[1] 陈凌燕[2] LIU Zhu-hua;CHEN Ling-yan(College of Physics and Communication Electronics,Jiangxi Normal University,Nanchang 330022,China;Department of Traditional Chinese Medicine,Jiangxi Maternal and Child Health Hospital,Nanchang 330006,China)

机构地区：[1]江西师范大学物理与通信电子学院,江西南昌330022 [2]江西省妇幼保健院中医科,江西南昌330006

出　　处：《计算机工程与设计》2020年第2期477-482,共6页Computer Engineering and Design

基　　金：国家自然科学基金项目（11764020）;江西省教育厅科研基金项目（GJJ160305）

摘　　要：针对超声波医学图像的斑点噪声滤除问题,提出一种保留细节的迭代非局部均值滤波方法。在非局部均值滤波基础上使用迭代操作,重新定义邻域相似块搜索范围的形式,使每个像素点的邻域相似块搜索范围随着迭代过程逐步增大,每一轮迭代的相似块搜索数量保持不变;引入一种边缘检测算子,对每轮迭代滤波前和滤波后的图像进行融合。实验结果表明,所提方法有更好的噪声滤除及细节保留效果,算法计算复杂度更低。Aiming at the de-speckling problem of ultrasonic medical image,a detail preserving iterative non-local means algorithm was proposed.Iterative operation was used on the basis of non-local means filtering,and the form of neighborhood range for searching similar blocks was redefined.It made the range for searching similar blocks of each pixel gradually increase with the iterative process,but the number of searches for similar blocks of each pixel in each iteration remained the same.An edge detection operator was introduced to fuse the images before and after filtering in each iteration.The experimental results indicate that the proposed algorithm has better effects on de-speckling and detail preserving,and it has lower computational complexity.

关 键 词：超声波医学图像 非局部均值 斑点噪声 迭代 图像去噪 保留细节 

分 类 号：TN911.73[电子电信—通信与信息系统]