Attacking “Photoshop Encryption”

Thomas Ptacek | January 6th, 2007 | Filed Under: Defenses, New Findings, Uncategorized

Dheera Venkatraman is clever. He proposes a way to extract sensitive information from “pixelated” or “blurred” portions of the pictures of checks, invoices, traffic tickets and whatnot that people post.

check.png

The observation: while you clearly can’t just “undo” the pixelation, if you know the blurring algorithm, you can iterate over candidate numbers (or images thereof) and see which ones produce the same image.

hash.png

It’s a simple dictionary attack (think of Photoshop’s “blur” as a cryptographic hash). This seems plausible because the “plaintext” is structured and relatively constrained, although in the real world you can think of a variety of things that would complicate the analysis (grainy source images, skew, etc). So I don’t know if it’s practical. But it made me think

Viewing 8 Comments

Trackbacks

close Reblog this comment
blog comments powered by Disqus