Comment **No Title** (Score 3) 413
Ross Anderson and a team of other researchers wrote a white paper entitled "On The Limits of Steganography" published in the IEEE Journal on Selected Areas in Communications Special Issue on Copyright & Privacy Protection, vol. 16 no. 4, pp 474-481, May 1998 (it's available online at http://www.cl.cam.ac.uk/~rja14/ ) that deals with the issue of robustness of watermarks and other forms of information hiding. The overall conclusions are that:
-Robustness decreases proportionally with the square of the information contained
-Watermarks can almost always be either distorted beyond recognition (if the information content is high) or removed (if the information content is lwo) using a simple sequence of transformations, ranging from smearing spectral power peaks to scaling the image
-It's almost always possible to determine that watermarks or steganography was used because the entropy of the bits affected is most probably higher than that of the surrounding message.
--
Flames? Think I'm a karma whore?
-Robustness decreases proportionally with the square of the information contained
-Watermarks can almost always be either distorted beyond recognition (if the information content is high) or removed (if the information content is lwo) using a simple sequence of transformations, ranging from smearing spectral power peaks to scaling the image
-It's almost always possible to determine that watermarks or steganography was used because the entropy of the bits affected is most probably higher than that of the surrounding message.
--
Flames? Think I'm a karma whore?
