Abstract: Image reconstruction from incomplete projections is a key problem in image reconstruction. In this paper, based on least-square optimization, some unknown error are adopted as inequality constraints. A new restricted iterative algorithm using Landweber iteration is proposed for the ill-posed problem. Compared with non-restricted iterative algorithm, under the condition of same computation time, this algorithm is more stable, and can give less error and higher fitness image. The effectiveness of the algorithm has been verified by image reconstruction using parallel-scanning data. The quality of reconstructed images clearly indicates the superiority of the proposed algorithm.