Sparse-view computed tomography (CT) iterative reconstruction suffers from severe image quality degradation, including reduced resolution and pronounced artifacts, due to insufficient projection data. As a result, it fails to meet the requirements for high-fidelity imaging in industrial and medical applications. To address this issue, this paper proposes a sparse-view CT iterative reconstruction algorithm that integrates multiscale nonlocal total variation and adaptive bilateral filtering (MA-SART). The proposed method extends the simultaneous algebraic reconstruction technique (SART) by incorporating a combined regularization strategy based on multiscale nonlocal total variation (MNLTV) and adaptive bilateral filtering (ABF) within the iterative reconstruction loop. This synergistic framework aims to achieve a globally optimal reconstruction while preserving fine details and effectively suppressing artifacts and noise. Specifically, the MNLTV term captures nonlocal similarities across multiple scales, thereby reducing gradient-related artifacts at the global structural level. In addition, the proposed ABF dynamically adjusts its spatial and intensity standard deviations and employs a gradient consistency constraint to modulate the range kernel weights, which suppresses artifacts near strong edges while preserving edge structures. The performance of MA-SART was validated using the Shepp–Logan phantom and wood transverse section datasets. Qualitative results show that images reconstructed using MA-SART closely resemble the reference images, particularly in regions with complex textures under magnification. For quantitative evaluation, the root mean square error (RMSE), peak signal-to-noise ratio (PSNR), and structural similarity index measure (SSIM) were used to compare MA-SART with ART, SART, SART-GSRRGF, and POCS-TVM. The experimental results demonstrate that MA-SART produces reconstructions that are closer to the original images and outperform the comparison methods in preserving fine structural details.
Li et al. (Mon,) studied this question.