Limited angle reconstruction with two dictionaries

In this work, a two-dictionary learning (TDL) based algorithm is proposed to solve the problem of image reconstruction in X-ray computed tomography (CT) with limited angle projections. One dictionary trained from a high quality image is used to improve the reconstruction image quality, and the other dictionary trained from artifacts is to reduce the limited angle artifact. Experiments with simulated projections and real data were performed to evaluate the proposed algorithm. The results reconstructed using the proposed method shows improved image quality compared with the reconstructions using an ART-TV method.