4 ms·
You need to quantify "blends in". If I [an attacker] plot a distribution of the entropy in all your files, and some of them are outliers, even by a small amoun
by blake8086 15y ago
You need to quantify "blends in". If I [an attacker] plot a distribution of the entropy in all your files, and some of them are outliers, even by a small amount, I can focus all my analysis on those files.
- waitwhat 15y agooutliers, even by a small amount Looks like you're getting the hang of it: good steganography relies on having better statistical models than the opposition. Things are weighted in favour of the concealer, however, as (unlike the attacker) he doesn't have to worry about false positives/negatives