利用h细化优化电阻层析成像灵敏度矩阵

Optimization of Electrical Resistance Tomography Sensitivity Matrix Using h Refinement

为了提高电阻层析成像图像重建质量,利用h细化优化灵敏度矩阵以改善其病态性.以电阻层析成像有限元模型h细化区域的起始层数、终止层数以及三角形有限元内部区域所插入节点的横坐标、纵坐标为变量,以敏感场均匀分布时灵敏度矩阵条件数的倒数为适应度函数,利用h细化优化灵敏度矩阵,并将优化结果应用于改进Landweber预迭代算法图像重建.实验结果表明,利用h细化可有效改善其病态程度,相比采取全局细化前后有限元模型对应的灵敏度矩阵,条件数分别降低了35.354,3%,、32.820,4%,,提高了重建图像分辨率.

In order to enhance image reconstruction quality,the sensitivity matrix was optimized using h refinement to cope with the ill-posedness in electrical resistance tomography.The starting and termination number of layer of the h re-finement region,the horizontal and vertical coordinates of the nodes were taken as variables,and the reciprocal of the con-dition number was designed as the fitness function,and the finite element mesh was optimized using h refinement,which was thereafter applied to image reconstruction using the improved Landweber pre-iteration algorithm.Experimental results show that the ill-posedness is improved effectively,the condition number is reduced by 35.354,3%, and 32.820,4%,comparing to the finite element mesh before and after global refinement,and the quality is improved.