当前位置: X-MOL 学术IEEE Sens. J. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
A Proportional Genetic Algorithm for Image Reconstruction of Static Electrical Impedance Tomography
IEEE Sensors Journal ( IF 4.3 ) Pub Date : 2020-12-15 , DOI: 10.1109/jsen.2020.3012544
Yijia Zhang , Huaijin Chen , Lu Yang , Kai Liu , Fang Li , Chen Bai , Hongtao Wu , Jiafeng Yao

A proportional Genetic Algorithm ( ${p}$ GA) is proposed to solve the inverse problem for the image reconstruction of static electrical impedance tomography (EIT). The ${p}$ GA obtains static EIT image with higher convergence speed and better reconstruction image quality by combining Genetic Algorithm (GA) with a ratio objective function. Although GA gets better performance for EIT inverse problem than any other traditional methods like Tikhonov regularization, it is high time complex and slow convergence. In this paper, a ratio objective function is proposed to solve the problem. Firstly, two kinds of methods: the Tikhonov regularization algorithm and the ${p}$ GA are used for image reconstruction in the simulation. The image quality of the two algorithms are compared. Secondly, the parameters such as population number and crossover mode of ${p}$ GA are optimized, initial values are set to improve the convergence speed of the algorithm, and the reconstructed image is processed to improve the quality of the image. Finally, experiments are conducted to verify the stability of the ${p}$ GA under certain noise conditions. In the experiment, different numbers and shapes of targets are placed in the sensor and an EIT system based on 34980A data acquisition system is used for data collection. The image reconstructed by ${p}$ GA shows that, the position and shape of the targets can accurately correspond to the real object.

中文翻译:

静态电阻抗断层成像图像重建的比例遗传算法

比例遗传算法( ${p}$ GA) 被提出来解决静态电阻抗断层扫描 (EIT) 图像重建的逆问题。这 ${p}$ GA通过将遗传算法(GA)与比率目标函数相结合,获得具有更高收敛速度和更好重建图像质量的静态EIT图像。尽管 GA 在 EIT 逆问题上比其他任何传统方法(如 Tikhonov 正则化)获得更好的性能,但它的时间复杂度高且收敛速度慢。本文提出了一个比率目标函数来解决这个问题。首先,两种方法:Tikhonov正则化算法和 ${p}$ GA用于模拟中的图像重建。比较两种算法的图像质量。其次,人口数量和交叉模式等参数 ${p}$ 对GA进行优化,设置初始值以提高算法的收敛速度,并对重构图像进行处理以提高图像质量。最后,进行实验以验证其稳定性 ${p}$ 特定噪声条件下的 GA。实验中,在传感器中放置不同数量和形状的目标,并使用基于34980A数据采集系统的EIT系统进行数据采集。重建的图像 ${p}$ 遗传算法表明,目标的位置和形状能够准确对应真实物体。
更新日期:2020-12-15
down
wechat
bug