A fast reconstruction strategy to image small objects in electrical tomography

Abstract

Reconstruction speed and spatial resolution are critical to image reconstruction in electrical tomography, but they are usually contradictory. To make a trade-off in the reconstruction of small objects, a fast reconstruction strategy based on two-stage Landweber method and block sparse Bayesian learning (BSBL) is proposed in this paper. First, an initial image is reconstructed by the two-stage Landweber method. Then, the region of interest (ROI) is determined by applying the Otsu thresholding. Finally, block sparse Bayesian learning is employed to reconstruct the targets in the region of interest. Numerical simulations were carried out to evaluate the performance of the proposed method. For comparison, several typical distributions containing multiple small objects of different sizes were reconstructed by the conventional and proposed methods. It is shown that the proposed method can provide more accurate images of small objects, and the average computational cost is less than 0.37 s.