Satellite images are the main source for different change detection applications. These images are usually taken from different sensors at different times. Hence, the point correspondence is not established. In this paper, the Singular Value Decomposition (SVD) is used to automatically solve the point correspondence problem between satellite images with different attributes. For each pair of the images, a cost matrix is built. The cost of corresponding any two points in a pair of images is computed using different geometric attributes. The SVD for the cost matrix is then used to solve the point correspondence problem. The algorithm is tested on QuickBird, IKONOS, and SPOT images. Results showed that using the traditional Euclidean distance cost matrix is not suitable for remote sensing images. Hence, a modified cost matrix was introduced based on the orientation parameters of the satellite images. The new cost matrix is then computed and its SVD is calculated. Results showed significant improvement in the solution of the point corresponding problem regardless the size or orientation of the satellite images.