To achieve four-phase image registration, an image registration method based on SIFT (scale invariant features transform) algorithm is proposed in this paper. The method is divided into four steps. Firstly, the feature points of the reference image and the misregistration images are extracted respectively. In this step, the characteristics of 2D-velocity interferometry system for any reflector (VISAR) images are fully considered, and homologous non-fringe images are introduced to obtain more accurate results. The second step is feature points matching. After roughly matching, a two-stepfiltering method composed of angle histograms and feature point distance is designed to achieve accurate matching. The third step involves calculating the transformation parameters based on the final matching results. Finally, the transformation parameters are applied to misregistration images for interpolation transformation to achieve image registration. One of the four-phase images is used as the reference image to register the remaining three images. For nonfringe images, experimental results show that the correlation coefficient between registered images and the reference image increases from 0.5 to above 0.9. For fringe images, the calculation accuracy of wrapped phase improves significantly. Therefore, the algorithm in this paper effectively solves the registration problem of 2D-VISAR four-phase images, laying a foundation for further data processing in the future.