The first image is easy, simply paste it into the top left corner of the composite image (as shown in red in the figure above). The next image, shown in blue, is more complex and needs to be rotated, scaled, and translated so that it correctly overlays the red image. The first step is to identify common feature points which are known as (). Then use the RANSAC tool to estimate the homography which maps 1p to 2p. Now map 2p to its corresponding coordinate in the first image Do this for every pixel in the new image by warping. As shown in the figure above, the warped blue image falls outside the bounds of the original blue image and the option 'full' specifies that the returned image is the minimum containing rectangle of the warped image. This image is referred to as a () and shown with a dashed black line. The vector t is returned by homwarp and gives the offset of the tile's coordinate frame with respect to the original image. In general, not every pixel in the tile has a (corresponding pixel) in the (original) image