Basically the opposite of the procedure I've described in a previous entry, and potentially (!) very useful. 'mutool extract paper.pdf' extracts all bitmaps in paper.pdf (as well as all fonts, which are not of interest here), and one thus gets the raw images without any of the artwork added to them for publication.
Here's a real-life example taken from this paper.
Unfortunately, this procedure only works if the publisher actually includes the submitted vector graphic files, and not screenshots of them. No, I'm not kidding. Using 'mutool' on several dozens of publications, I found that many (if not most) publishers actually convert the submitted eps files into bitmaps, including low-quality jpegs.
Now, you can call these publishers ignorants (and rightly so), or even cretins, dipshits, imbeciles, idiots, morons, or whatever qualification you would like to emphasize. But remember that we usually pay them handsomely to get our work published in their journals. Now who's the idiot?
'mutool' is part of mupdf, in case you were wondering.