Order subset expectation maximization (OSEM) iterative algorithm is rapidly developed in recent years. But this algorithm is easy to generate some striation artifact, metal artifact or scattering artifact during iterative process. So a new iterative method named smooth constrained OSEM (SC-OSEM) is built, which utilizes smoothing constrained matrix as priori information in OSEM reconstruction. Median filtering algorithm and total variation minimization algorithm (TVM) are introduced as smoothing constrained conditions. The images reconstructed by computer tomography are consistent well with the initial model for the situations of incomplete projection data with ideal, metal, and noise, whose reconstruction qualities are better than only OSEM iterative algorithm. Median filtering constraint makes reconstructed image denoised, while TVM makes metal boundary clearer. These results manifest that SC-OSEM iterative algorithm is an adaptable and high precision CT reconstruction method.