Skip to main content
4 events
when toggle format what by license comment
Aug 31, 2018 at 14:15 comment added Cort Ammon Here's an article, which links a paper. They were able to use hidden markov models to substantially automate the process. They presented success rates against various kinds of text (account numbers, natural language, etc.). That being said, I agree fully that the pixelation method does look better.
Aug 31, 2018 at 9:43 comment added IconDaemon I’ve found the Lifehacker article that posits this ‘method’ of deobfuscation, but I’m not convinced of its practicality. I favor the pixillation method because it has a softer look than the ‘redacted’ appearance one finds in declassified secret documents.
Aug 31, 2018 at 2:58 comment added Cort Ammon Worth noting is that there are known methods to recover information hidden behind pixelated blocks like this. Know your threat model: what kind of adversary are you worried about? In particular, this sort of attack has been used against people who pixelate their account number on checks. The format of the numbers on a check is very regular, making the attack easier. Arbitrary text is harder to recover than that.
Aug 30, 2018 at 19:30 history answered IconDaemon CC BY-SA 4.0