From the description on the link you posted:
Use this method if you have an image that you want to place in a scene, but prefer to have only a single image as a resultant. The following example achieves this by replacing the scene pixels with the object image pixels.
So I'm guessing it is doing what it is meant to do! but if you want the underlying pixels to be shown too, you need to edit:
result((1:size(obj,1))+rowshift, (1:size(obj,2))+colshift, :) = 0.25.*obj + 0.25.*scene((1:size(obj,1))+rowshift, (1:size(obj,2))+colshift, :);
in the part where result is assigned the pixel values. Note that the 0.25 scaling is just so that the index values don't go > 1.
Hope this helps!
