Regularization parameters determination method based on random matrix clustering

Electrical Impedance Tomography (EIT) is an advanced visualizing technique with non-radiation, fast-response, and non-invasive characteristics, and has been widely used in medical monitoring, industry detection, biology engineering, etc. However, the low spatial resolution of EIT technique limits its development in the application fields. The regularization method is a typical algorithm to improve the reconstruction quality of EIT image. And its regularization parameter is the key for it to work effectively. However, the existing methods to determine regularization parameter cannot work effectively in many cases. In order to improve the quality of the EIT image, in this paper, a random matrix-based clustering algorithm is proposed to determine the key regularization parameters and then uses the Tikhonov Regularization algorithm for EIT image reconstruction. The results show that based on the clustering algorithm, better regularization parameters can be determined and the quality of the reconstructed image can be improved.