The optimization of electrical tomography algorithms

The paper discusses the optimization of methods for fast image reconstruction in electrical resistive/impedance/capacitance/infrared tomography. The first portion of the text characterizes the total variation and Tikhonov regularization techniques, including their advantages and drawbacks. Within the related other part, the authors then focus on time optimization in computing the distribution of the desired quantity inside a given object. In this context, the following options are considered: a) adaptive control of the regularizing element of the objective function; b) parallelizing the computation of the Jacobian and the Gauss-Newton system of equations; c) compressive sensing. The parallelization of the algorithm was outlined with respect to the capabilities of a general purpose GPU. In terms of its general goal, the research is intended to design universal libraries for reconstructing complex quantities inside measured objects.