A novel image restoration algorithm is presented in this paper. The restoration problem is formulated under the energy minimization framework and is solved using a dynamic programming-based approach. Through applying dynamic programming iteratively along both horizontal and vertical scanlines, the new algorithm can quickly converge to a near-globaloptimal solution, without suffering the so-called streak artifacts. Experiments on both grayscale and color images demonstrate that the presented algorithm can effectively remove Gaussian noise and impulse noise from corrupted images, as well as to restore images with missing intensity values.