基于MLEM-TVM算法的层析γ扫描图像反演方法研究  

Image Inversion of Tomographic Gamma Scanning based on MLEM-TVM Algorithm

作　　者：郑洪龙 冯昊筠 庹先国 石强 韩强 刘崎 牟煜璇 王洲 苟睿 李宇浩 ZHENG Hong-long;FENG Hao-jun;TUO Xian-guo;SHI Qiang;HAN Qiang;LIU Qi;MOU Yu-xuan;WANG Zhou;GOU Rui;LI Yu-hao(Sichuan University of Science&Engineering,Zigong Sichuan 643000,China;Nuclear Power Institute of China,Chengdu 610005,China)

机构地区：[1]四川轻化工大学,四川自贡643000 [2]中国核动力研究设计院,成都610005

出　　处：《核电子学与探测技术》2025年第2期162-171,共10页Nuclear Electronics & Detection Technology

基　　金：国家自然科学基金(42204179,42374227);四川省科技计划项目(2024NSFSC1358);四川轻化工大学研究生创新基金(Y2024300);核测量方法与智能装备“652”科研创新团队(SUSE652A001)资助。

摘　　要：层析γ扫描技术(Tomographic Gamma Scanning,TGS)是目前最为先进的桶装放射性废物γ无损检测技术之一,对于投影数据欠采样条件下的TGS图像反演,传统迭代算法不再适用。本文将基于压缩感知理论的MLEM-TVM算法应用于废物桶TGS图像反演中,在传统MLEM算法迭代的基础上增加TVM迭代过程,提高投影数据欠采样条件下的TGS图像反演的准确性。结合阵列碲锌镉探测器的TGS系统、桶内介质与核素,利用蒙特卡罗程序MCNP5开展废物桶TGS扫描实验模拟,分别采用MLEM算法和MLEM-TVM算法反演桶内介质线衰减系数分布图像和核素活度分布图像,结果表明:MLEM-TVM算法反演桶内介质线衰减系数分布图像的均方差约为MLEM算法的0.18倍,信噪比约为MLEM算法的2.19倍,同时MLEM-TVM算法反演桶内核素^(137)Cs、^(60)Co总活度值也比MLEM算法更加接近实际值。该方法在兼具较高的TGS检测效率、较好的核素识别能力和成像质量方面具有一定参考意义。Tomographic gamma scanning(TGS)is currently one of the most advanced techniques for gamma non-destructive assay of radioactive waste drum.For the inversion of TGS image with undersampling projection data,traditional iterative algorithms are no longer applicable.In this work,the MLEM-TVM algorithm based on the theory of compressed perception is applied to the inversion of TGS image.The TVM iteration process is added to the traditional MLEM algorithm to improve the accuracy of TGS image inversion with undersampling projection data.Tomographic gamma scanning for waste drum is simulated with the Monte Carlo program MCNP5 in conjunction with the TGS system of arrayed CdZnTe detectors,medium and nuclides in the drum.The MLEM algorithm and the MLEM-TVM algorithm are used to invert the image of the linear attenuation coefficient distribution and the image of nuclide activity distribution in the drum,respectively.The result shows that the MSE of the image of linear attenuation coefficient distribution in MLEM-TVM algorithm is about 0.18 times higher than that of the MLEM algorithm,the SNR is about 2.19 times higher than that of the MLEM algorithm,and the values of the activity of nuclides ^(137)Cs and ^(60)Co in the drum are much closer to the actual values than that of the MLEM algorithm.The method is informative in terms of combining high TGS detection efficiency,better nuclide identification and imaging quality.

关 键 词：放射性废物桶 层析γ扫描 迭代算法 图像反演 

分 类 号：TL814[核科学技术—核技术及应用]