warp one of the images (called the “moving image”) over all possible angles of rotation and measure similarity between this image and the other one which is kept fixed (called “reference image”)