ABSTRACT
ABSTRACT: In Region-Of-Interest(ROI) tomography, image reconstruction is an ill-conditioned inverse problem due to the presence of additive noise and incomplete projection data. To improve reconstructed image quality, this paper is aimed at reconstructing a small portion of an object from noisy observations of its projections sampled in and near the ROI. The proposed method can divided into two steps to reconstruct ROI of an object from a set of its noisy projection lines that passed through ROI. First, we introduce filter matrices to filter noisy projection data in multiresolution local tomography. Second, we apply total variation(TV) methods to make reconstructed image deblurring. Experimental results show that the proposed method outperforms state-of-the-art method.
