Document and image management systems have been doing this sort of thing for decades. Form a unique identifier for an image by hashing, split the bytes out into a path name by hex converting, and you have a file system path.
I'd say that this particular troll is going to get stepped on pretty hard, and all of these patents are going to be invalidated.
I never claimed I invented it ;-) It just seemed like the obvious way to prevent duplication when someone would rather update the same file again than search for a previously existing version.
In the end, when a dupe was uploaded for the second time, the uploader would be rewarded by not having to edit any metadata - as they would be redirected to the original asset page.
If someone ever uploads a second, valid, JPEG that collides in md5-space with a previously uploaded one, the sysadmins in the app group will receive an e-mail commemorating the fact someone defeated our deduper.
I'd say that this particular troll is going to get stepped on pretty hard, and all of these patents are going to be invalidated.