The Research Of Electrical Impedance Tomography Image Reconstruction Algorithm Based On Non-uniform Reference Field

During recent decades, the Electrical Impedance Tomography technology is developing fast because of its safety, fast speed, low cost and non-invasive property, it attracts more and more attention from scientists and engineerings all over the world, and it has been applied to industrial and clinical area.In most studies, EIT image is reconstructed based on uniform reference field, however, in the practical application such as multi-phase flow measurement or biomedical field, it’s difficult to achieve measurement data of uniform reference field. In this paper, the authors use the non-uniform reference field for research. New methods including conductivity fitting method and gradient method have been put forward in order to solve the image reconstruction problem, the reconstructed image’s quality is improved.A reference conductivity fitting method is created to calculate quasi-absolute conductivity image, which equals the product of quasi-reference conductivity and relative image, while the reference value is derived from the comparison between different fields’ boundary voltage. In the MPF measurement, gas/liquid flow’s void fraction with higher accuracy can be deduced from the absolute image, which indicates the improvement of the image quality. In addition, a Gradient Method used to identify object boundary in EIT image has been put forward, which contributes to reconstruct image of higher accuracy within the object region. Parts of simulation and experiment based on non-uniform reference field will be demonstrated in the paper, both forward problem and inverse School of Electronic Information and Electrical Engineering....................................... 2 problem of EIT are solved; particular criterias are introduced to evaluate these new methods’ correctness and accuracy compared with the traditional methods.Reference conductivity fitting method is a simple and effective method, it doesn’t need the background conductivity, the achieved quasi-reference conductivity is close to average conductivity, it can be used to reconstruct quasi-absolute conductivity image which is close to reality in investigated situations, moreover, the reconstructed quasi-absolute conductivity image has the same level of relative error compared with the traditional method that uses continuous phase’s conductivity as reference conductivity, so it can be applied widely especially when background conductivity cannot be achieved. The void fraction of gas/liquid flow with much higher accuracy can be deduced by reference conductivity fitting method, which proves that the new method can increase the image reconstruction quality.Both the simulation and experiment results show that Gradient Method can identify the object boundary accurately, the estimated boundary is able to cover the object completely without prior knowledge, while the traditional threshold method largely relies on the definition of threshold value, when the value is proper, the boundary accuracy will be high, but when it is not, the accuracy will be low, or even worse, the estimated boundary cannot cover the whole anomalies which may cause serious consequence. The gradient method can help achieve reconstructed image with higher accuracy within the object region.The study in this paper helps improve the EIT applications in both industrial and clinical area.