Abstract: To improve image reconstruction quality of magnetic induction tomography (MIT)， a new MIT image reconstruction method based on compressed sensing was put forward. Here the voltage collection process of MIT system was regarded as linear measurement process of compressed sensing， the measurement matrix was designed by zero expansion of the sensitivity matrix and its row vectors random reconstruction， and the measurement voltage vector expanded by the same way was regarded as measurement signals of compressed sensing. The original signal was recovered by signal reconstruction method of compressed sensing. Finally， the simulation experiments were performed. The experimental results indicated that the average image reconstruction error and the average correlation coefficient obtained by the method are better than corresponding indicators obtained by the traditional reconstruction image algorithm. So it is a kind of MIT image reconstruction method with high accuracy.

Key words: magnetic induction tomography(MIT), compressed sensing, image reconstruction, image error, correlation coefficient