中文摘要：

为了改善电阻抗图像重建质量,提出阈值修正法,并验证该方法的有效性。首先,阐述阈值修正法原理及阈值选取原则;其次,采用高斯牛顿算法和等位线滤波反投影算法,在多目标分布情况下,合理选取阈值,修正电导率,重构电阻抗图像;然后,对修正前后图像的相对误差和相关系数进行对比分析;最后,对物理模型实测数据进行成像,并分析修正前后的两项图像性能指标。仿真结果显示,这两种算法在各自合理阈值的修正下,修正后重构图像的相对误差比修正前的相对误差分别平均减小了45.68%和48.36%,相关系数平均提高了33.86%和50.86%。实测结果显示,修正后的相对误差平均减小了47.69%,相关系数平均提高了11.26%。采用阈值修正法,合理选取阈值,可显著去除图像伪影,改善成像质量。该方法有效可行,有利于EIT技术在工业及生物医学工程领域发挥更大的优势。

英文摘要：

In this paper,a threshold correction method was proposed to improve the quality of electrical impedance images,and efficiency of it was tested and verified.The images were reconstructed by Gauss-Newton algorithm and equi-potential filter back-projection algorithm respectively for 1 to 7 objects distributed within a circle field.The images were evaluated by the relative error and the correlation coefficient performance targets.Testing data were used as well to verify the efficiency of the proposed method.The simulation results showed that the average relative errors of corrected images decreased by 45.68% and 48.36%,and the average correlation coefficients increased by 33.86% and 50.86% according to the adopted algorithms when reasonable threshold values were taken.The experimental results showed that the average relative error of corrected images decreased by 47.69%,correlation coefficient increased by 11.26%.The proposed method is effective,feasible to improve reconstruction quality of EIT images in inhomogeneous field.