When you invoke the dialog to insert a link after you selected a link in the wikitext, the fields in the dialog are filled accordingly. The same would be nice for the image dialog, too. This is not possible (or at least quite difficult) for some cases, since there are much more possible keywords and localizations, but in those cases where only known keywords are used, this should be possible.
An example use case: A user adds an image using the dialog, previews the page and decides to move the image to the left side. Then he should be able to select the code for the image, invoke the dialog again and just change the position.