
电子设计工程
Electronic Design Engineering
第 26卷
Vol.26
第 5期
No.5
2018年 3月
Mar. 2018
收稿日期:2017-03-10 稿件编号:201703122
基金项目:国家自然科学基金资助项目(11261061,61362039,10661010);新疆师范大学研究生科技创新项目
资助(XSY201602010)
作者简介:唐 泉(1991—),男,河南新蔡人,硕士研究生。研究方向:偏微分方程数值解法。
乳腺癌是女性中主要癌症类型,也是女性中第
二大致死性癌症类型,早期检测和治疗被视为最有
效的方法来降低乳腺癌的死亡率。它们通常由于图
像质量差,普遍缺乏对比度,存在伪影、模糊、噪声和
阴影等。虽然目前已经有好多学者提出医学图像增
基于二进小波与 PDE 的乳腺 X 线图像增强
唐 泉,黄允浒
(新疆师范大学 数学科学学院,新疆 乌鲁木齐 830017)
摘要:腺钼靶 X 线摄影成像是乳腺癌早期发现、早期诊断的最有效方法,具有重要价值.本文提出结
合à trous 算法与偏微分方程的乳腺 X 线图像增强,该方法在二进小波理论基础上,利用推广的二维
à trous 算法对乳腺图进行分解,接着根据低频和高频系数特点设计了不同的软硬阈值法对经过分
解后的低频子带进行去噪预处理,进而利用最大类间方差法的 7 段对称连续分段函数增强;而高频
在偏微分方程的 P-M 正则化模型基础上利用一个新的扩散系数,调整常数α以达到对高频子带噪
声较好抑制,后用改进的具有动态范围压缩和颜色保真的多尺度 Retinex 算法进行增强;最后利用
à trous 算法重构增强后的低频和高频子带.为验证该方法的有效性,我们还将其与传统的图像增强
方法作比较;仿真试验结果表明,该方法对乳腺 X 线图像的增强效果远远好于传统的增强方法。
关键词:二进小波;偏微分方程;二维à trous 算法;分解与重构;多尺度 Retinex 算法;Otsu 算法
中图分类号:TN97 文献标识码:A 文章编号:1674-6236(2018)05-0022-05
Breast X - ray image enhancement based on dyadic wavelet and PDE
TANG Quan,HUANG Yun⁃hu
(College of Mathematical Sciences,Xinjiang Normal University,Urumqi 830017,China)
Abstract: Gland mammography imaging is the most effective method for early diagnosis of breast cancer
early detection,and has important value. This paper proposes à trous algorithm based on mammograms
and partial differential equation method in the enhancement,dyadic wavelet theory based on the
promotion of the use of a 2- D à trous algorithm to decompose the breast,then according to the low
frequency and high frequency coefficients of the design characteristics of different soft and hard threshold
method on the low frequency subband coefficients are denoised,then using the Otsu method 7 symmetric
continuous piecewise function enhancement;high frequency by using a new diffusion coefficient based
on P- M regularization model of partial differential equations. In order to achieve the constant α
adjustment of the high frequency subband noise restrain,enhancement of multi-scale Retinex algorithm
with improved dynamic range compression and color fidelity;finally,Finally,the à trous algorithm is
used to reconstruct the enhanced low frequency and high frequency sub bands. To verify the validity of
the method,we have with the traditional image enhancement methods are compared;simulation results
show that the enhancement effect of breast X- ray image enhancement is far better than traditional
methods.
Key words: dyadic wavelet;partial differential equation;2- D à trous algorithm;decomposition and
reconstruction;multi scale algorithm;Otsu method
-- 22