计算物理 ›› 2003, Vol. 20 ›› Issue (5): 439-442.

• 论文 • 上一篇    下一篇

基于径向基函数神经网络的线性衰减系数的重建算法

张全虎1,2, 张其欣1, 吕峰1, 李泽1, 李鲲鹏1, 王仲奇1, 赵学军1, 隋洪志1   

  1. 1. 中国原子能研究院, 北京 102413;
    2. 西安市第二炮兵工程学院, 陕西 西安 710025
  • 收稿日期:2002-05-29 修回日期:2002-12-29 出版日期:2003-09-25 发布日期:2003-09-25
  • 作者简介:张全虎(1965-),男,山西洪洞,博士生,从事原于核物理及棱保障技术方面的研究,北京275信箱48分箱.

Reconstruction Algorithm of Radial Function Neural Networks for Linear Attenuation Coefficients

ZHANG Quan-hu1,2, ZHANG Qi-xin1, LÜ Feng1, LI Ze1, LI Kun-peng1, WANG Zhong-qi1, ZHAO Xue-jun1, SUI Hong-zhi1   

  1. 1. China Institute of Atomic Energy, Beijing 102413, China;
    2. The Second Artillery Engineering College, Xi'an 710025, China
  • Received:2002-05-29 Revised:2002-12-29 Online:2003-09-25 Published:2003-09-25

摘要: 层析γ扫描(TGS)技术是非破坏性分析(NDA)中的一项重要技术.在TGS透射测量中,线性衰减系数值的图像重建问题是TGS的难点和核心问题.在文[1]的基础上,提出了将神经网络方法应用于TGS重建线性衰减系数图像的算法.计算机上的仿真模拟结果表明,在一定范围内,径向基函数(RBF)神经网络方法重建的线性衰减系数值与实际值的相对误差小于4%,且具有快速、高精度等优点,表明了此方法的有效性.

关键词: 层析γ扫描, 线性衰减系数, 径向基函数, 神经网络, 图像重建

Abstract: Tomographic Gamma scanning(TGS) technique is an important technique in nondestructive assay(NDA). The image reconstruction problem of linear attenuation coefficients is very difficult and central in transmission TGS. A new image reconstruction algorithm of linear attenuation coefficients with neural networks is proposed based upon paper [1]. Simulated results indicate that the reconstruction relative errors of linear attenuation coefficients are less than 4% for the reconstruction algorithm of radial basis function(RBF) neural networks and the new algorithm has the merits such as fast response and high accuracy within a certain scope.

Key words: tomographic Gamma scanning, linear attenuation coefficient, radial basis function, neural networks, image reconstruction

中图分类号: