基于剪切波变换的辐射图像泊松噪声降噪技术研究  

Denoising Technology for Radiation Image with Poisson Noise Based on Shearlet Transform

作　　者：许玉婷 吴志芳[1,2] 王强 侯永明 赵斌 刘欣侠 XU Yuting;WU Zhifang;WANG Qiang;HOU Yongming;ZHAO Bin;LIU Xinxia(Institute of Nuclear and New Energy Technology,Tsinghua University,Beijing 100084,China;Beijing Key Laboratory on Nuclear Detection and Measurement Technology,Beijing 100084,China;Chinese Academy of Customs Administration,Qinhuangdao 066004,China;School of Vehicle and Energy,Yanshan University,Qinhuangdao 066004,China)

机构地区：[1]清华大学核能与新能源技术研究院,北京100084 [2]核检测技术北京市重点实验室,北京100084 [3]中国海关管理干部学院,河北秦皇岛066004 [4]燕山大学车辆与能源学院,河北秦皇岛066004

出　　处：《原子能科学技术》2022年第3期577-584,共8页Atomic Energy Science and Technology

基　　金：秦皇岛市市级科学技术研究与发展计划项目(202005A002)。

摘　　要：为了降低由统计涨落引起的辐射图像噪声,提出了一种基于剪切波变换的降噪方法。该方法以低剂量射线或质量厚度大的物体的辐射图像为研究对象,对此类辐射图像进行了噪声分析,利用Anscombe变换将统计涨落引起的泊松噪声转换为高斯噪声,再运用剪切波分解、阈值去噪、剪切波重构和Anscombe逆变换得到降噪图像。结果表明,当剪切波分解层数为5,采用改进阈值函数及极小极大原理阈值时可达到最优降噪效果,该方法能较好地去除辐射图像中的泊松噪声并保留边缘、细节信息,在视觉和量化指标上均优于传统降噪方法。In order to reduce the radiation image noise caused by statistical fluctuation,a denoising method based on shearlet transform was proposed.The radiation image of low-dose radiation or object with large mass thickness was taken as research objects.Through noise analysis,Anscombe transform was used to convert Poisson noise caused by statistical fluctuation into Gaussian noise,then shearlet decomposition,threshold denoising,shearlet reconstruction and Anscombe inverse transform were utilized to obtain the denoised image.The results show that the optimal denoising effect can be achieved when the scale of shearlet decomposition is 5 and the improved thresholding and the threshold of minimax principle are chosen.This method can reduce Poisson noise and retain image details.Moreover,it is superior to the traditional methods in both visual feeling and quantitative parameter.

关 键 词：辐射成像 泊松噪声 降噪技术 剪切波 Anscombe变换 

分 类 号：TL822.6[核科学技术—核技术及应用] TP751.1[自动化与计算机技术—检测技术与自动化装置]