This paper presents a new method for obtaining a rotation and translation-invariant scale signature of an image, a scale and translation-invariant rotation signature, and a technique for recovering the rotation, translation and scale transformation parameters that relate two images of similar scenes using these signatures. The transformation parameters can be used to coregister the subbands of a composite image or to register time-series images for further analysis, common tasks in both satellite imagery and radiology research. In the nite discrete case where the invariances do not hold completely, compared to previous techniques the technique is extremely robust in the presence of occlusion due to alignment (image data moving in and out of the frame), resampling, linear and constant luminance changes, and noise. This robustness comes from the use of filters that minimize transformation artifacts in the Fourier-Mellin domain.