基于分裂Bregman方法的加权频差电阻抗成像算法  

Weighted frequency difference electrical impedance tomography algorithm based on split Bregman method

作　　者：成民民 戎舟[1] 庞宗强[1] Cheng Minmin;Rong Zhou;Pang Zongqiang(Automation College, Nanjing University of Posts and Telecommunications, Nanjing 210000, China)

机构地区：[1]南京邮电大学自动化学院,南京210000

出　　处：《国外电子测量技术》2019年第2期30-35,共6页Foreign Electronic Measurement Technology

基　　金：国家自然科学基金青年科学基金(11604158)项目资助

摘　　要：加权频差阻尼最小二乘算法引入正则化方法,对病态的敏感矩阵进行修正,克服了背景对成像结果的影响,在背景阻抗不随频率变化时能够反映异物的位置,但当背景的阻抗随着频率的变化而变化时,背景区域的伪影较多,掩盖了异物区域。由于加权频差中,重建两种频率下的阻抗差值具有稀疏性,针对这一特性,引入L1-范数进行约束并采用分裂Bregman方法对其进行优化求解,改善了这一问题。实验结果表明,优化后的加权频差算法能够减少成像中存在的伪影并能进一步提高成像质量。因此,改进的加权频差算法是一种有效的准静态电阻抗成像算法。The weighted frequency difference damped least squares algorithm introduces a regularization method to correct the ill-conditioned sensitivity matrix, which overcomes the influence of background on the imaging results. It can reflect the position of foreign objects when the background impedance does not change with frequency.However, there are more artifacts in the background areawhich masks the foreign matter areawhen the background impedance changes with the frequency. Because of the sparsity of the impedance difference between the two frequencies in the weighted frequency difference, this paper improves the problem by introducing the Ll-norm constraint and using the split Bregman method to solve the problem. The experimental results show that the optimized weighted frequency difference algorithm can reduce the artifacts existing in the imaging and further improve the imaging quality. Therefore, the improved weighted frequency difference algorithm is an effective quasi-static electrical impedance imaging algorithm.

关 键 词：加权频差 分裂Bregman方法 稀疏性 L1范数 

分 类 号：R318[医药卫生—生物医学工程]