Legal
The views expressed in this blog are my own and do not necessarily reflect the views of Adobe Systems Incorporated.
Search
January 8, 2007
Decoding scrambled pixels
Removing data from a digital file is sometimes easier said than done. Redacting PDFs has sometimes proven tricky (something the Acrobat team has worked to address), and now a research report notes methods for unscrambling numbers or text that have been obscured via simple Photoshop tricks. [Via]
I asked a few Photoshop engieneers for comments & got some useful nuggets:
- Gregg Wilensky says, "The ability to recover text that is blurred is limited by the amount of noise in the image (and knowledge of the blurring function). So, adding a bunch of noise to the image is better, but still not foolproof. I would suggest completely replacing the text with noise and blurring that (for aesthetics)."
- Jerry Harris notes, "Having a known set of limited targets, OCR numbers in his example, makes the reconstruction task a bit more realistic in terms of useful results than most."
- And Todor Georgiev writes, "Using a known set of blur kernels (those in Photoshop!),
and a known set of targets, limits the set of possible outcomes
and makes this technique work. But slightly change lighting
and/or use custom blur filter, and your data is safe."
Comments