2014 , Volume 19, № 2, p.62-75

Two methods are developed to solve the following problem of ROI-tomography: the region of interest is placed in the central part of the image, while the projection data contain noise that has its maximal intensity on the periphery. In one of these methods, a multiplication of the projections by a function of window is applied. Using another method, one divides each projection into several parts, which are separately filtered and multiplied by the weights, and then the obtained functions are summarized. The estimations of noise component amplification are derived. Numerical simulations show that a classical approach, which consists in the immediate implementation of the inverse Radon transform formula, provides much faster growing of the reconstruction error when the noise level increases compared to the proposed methods.