In order to reconstruct high-quality image from projection data at few views, we develop an improved fast iterative reconstruction algorithm based on the minimization of the image total variation (TV). In the algorithm, conjugate gradient method is applied to solve the minimization of the image total variation, and multi-resolution iterative technique is used in the iteration process. We have performed numerical experiments using both computer-simulation data and real CT data. Experimental results demonstrate that the proposed algorithm not only improves the quality of image reconstructed, but also significantly increases the convergence speed of iteration image.