Referring to the problems of Synthetic Aperture Radar (SAR) image registration, an approach of SAR image registration based on affine invariant Scale Invariant Feature Transform (SIFT) features is proposed. First, the affine invariant SIFT descriptors are constructed by improving the original SIFT descriptors and the control points are rough matched by using the improved SIFT descriptors. Then, a new proximity matrix is built according to the scale ratio and the difference of orientations among the rough matched points and the similarity of the intensity around the rough matched points. Finally, the exact matching points are determined by using Singular Value Decomposition (SVD) method, and the precise image registration is performed. Experimental results show that the proposed algorithm has a better performance than original SIFT algorithm and SIFT+SVD method, and the accuracy of our algorithm can achieve sub-pixel level.