Re: Government proposes email and internet tracking


"Cynic" <cynic_999@xxxxxxxxxxx> wrote in message
news:40qia4pdh508rneu5q80008ocs4e76t4c7@xxxxxxxxxx
On Mon, 18 Aug 2008 10:34:02 +0100, MM <kylix_is@xxxxxxxxxxx> wrote:

Just possibly the statistics of the replaced bit might give it away -
unless you're careful, the distribution of bits in the hidden data
(presumably encrypted) would be different from those expected for a
picture.

Well, yes, you might have a run of 1's or 0's, but so might you have
in the data you're wishing to encrypt.

The point is the other way around. The LSb of a CCD pixel is pretty
much random (unless it is completely dark or fully saturated).


You need to be careful with terminology here. The LSB of a pixel from a CCD
will never be random as it is determined by the illumination that the CCD
element receives from the subject and not by the operation of chance.
Selecting a number of pixels at random, the LSBs may themselves be random.
However, comparing adjacent pixels, the LSB will be determined by the
characteristics of the light landing on the CCD sensor, and for a single
feature of the image stand a chance (maybe small but certainly not zero) of
being the same or nearly the same (with a small variation due to the noise
characteristics of the sensor itself - if you chose a bright subject with a
camera with a large low sensitivity sensor this will also deviate from what
seems to random).

So I would expect a completely random distribution over the entire
image - which means almost the same number of 1's and 0's.


You might have the same number of 1's and 0's but they will not be randomly
distributed as they are not determined by pure chance.

If the data you are encrypting does not have an equal number of 1's
and 0's, the distribution of the whole image will be skewed and give
the game away.


Which is more or less how steganographic detection software works - or at
least that is the starting point.



.

Relevant Pages

  • Re: Probability density of edges between clusters
    ... I do not want to describe the clusters, I want to derive a probability ... and calculate binomial p for each pixel. ... neighbourhood around each pixel and define the probability of an edge ... Ideally it is a binomial distribution per pixel. ...
    (sci.stat.math)
  • Re: Generate noise model[Please help]
    ... getting random numbers taken from whatever distribution you want. ... What you have not given is how that noise relates to the pixel value. ... where all the Rayleigh noise comes from the same ...
    (comp.soft-sys.matlab)
  • Re: sampling from a circular distribution
    ... > distribution is roughly uniform, ... The distribution of this sample is even less uniform. ... If you only have 180 possible pixel values (did I understand your ... standard chi-square test for significance since you know the observed ...
    (sci.stat.math)
  • Re: sampling from a circular distribution
    ... The distribution of this sample is even less uniform. ... > standard chi-square test for significance since you know the observed ... > and expected values for each pixel value. ...
    (sci.stat.math)
  • Re: Two booleans out of one number?
    ... Depending on the range (and distribution) of the number the below might ... Note that checking if the MSB ... and LSB is odd or if they are above the middle of thier range does ...
    (comp.programming)