Most pose estimation algorithms fail to deal with pose ambiguities when using coplanar feature points. Existing approaches which have investigated pose ambiguities suffer from either low performance or heavy computational cost. In this paper, we present a fast and robust scheme for camera pose estimation using coplanar feature points. First, the two local minima of pose under paraperspective projection are computed for coarse estimations. Then, these two poses are refined using the Orthogonal Iteration pose estimation algorithm to get the two local minima under perspective projection. Finally, the minimum with the lower object collinearity error is chosen for the correct pose. Experimental results with synthetic camera images validate the accuracy and robustness of the proposed algorithm, and show that its computational complexity is much lower than the state-of-the-art approach.