In the article, an iterative reconstruction algorithm based on total variation minimization and POCS optimization for non-Cartesian K-space data is proposed. The proposed algorithm interpolates non-Cartesian data onto a 2D Cartesian grid using gridding method first, and then during the iterative process of total variation minimization, the frequency values on grid points near the measured data are replaced with the interpolated ones according to POCS. The experiments on simulated and real data show that the proposed method can reconstruct image more accurately and rapidly than constrained total variation minimization method.