Jalil Taghizadeh-Tameh; Hossein Mousazadeh; Shahin Rafiee; Nazilla Tarabi
Abstract
Magnetic induction tomography (MIT) is one of the new imaging techniques, and due to its characteristic such as non-intrusive and non-destructive, it has potential for use in many industries, including biological industries, multiphase flows, medical imaging, agriculture and food industries. One of the ...
Read More
Magnetic induction tomography (MIT) is one of the new imaging techniques, and due to its characteristic such as non-intrusive and non-destructive, it has potential for use in many industries, including biological industries, multiphase flows, medical imaging, agriculture and food industries. One of the main parts of magnetic induction tomography system is the inverse problem solution algorithm. In this research, performance comparison of four image reconstruction algorithms in applied current magnetic induction tomography (AC-MIT) system was investigated. This system has two innovative annular electrodes as transmitter sensors and 648 coils as a receiver sensor. In order to evaluate the system performance, 12 combinations of target objects were used and image reconstruction was performed using linear back projection algorithm, Landweber iterative algorithm, Tikhonov regularization method and iterative Gauss-Newton algorithm. Size error (SE) and Relative image error (IE) parameters were used to evaluate the quality of the reconstructed images. The results showed that in all combinations of target objects, the IE values of iterative Gauss-Newton algorithm are lower than other algorithms. The results of size error parameter showed that in all four image reconstruction algorithms; increasing number of target objects increases SE parameter. As a general conclusion, it can be stated that iterative Gauss-Newton algorithm has a better performance compared to other algorithms.
