Revision history [back]

One method of doing so would be to detect keypoints in the two images, extract descriptors and match them. Then, you can take the bounding box of the matched keypoints in the two images.