Cryptography Reference
In-Depth Information
Figure 9.2: These are the least significant bits of the photo in Figure
9.1. The most significant bits were deleted to show the randomness
that exists at this level. (See Figure 9.4.)
out because a pure tone is converted to one with a bit of noise. 1
“Putting JPEG to Use”
on page 183 shows how
compression can
identify just how much
space can be exploited
in an image.
9.2.3 Independence Problems
One of the deeper problems is defining good noise. The least signifi-
cant bits of a music or image file often seem close to random, at least
to the average eye or ear, but they often contain hidden patterns and
structure. Many of the microphones, cameras or scanners used to
generate the files are far from perfect and they often introduce their
own patterns.
One of the most common is a correlation between the high order
bits and the least significant bit. A picture of a bright day might
include a number of intense reflections of the sun. These pure white
points usually saturate the sensor and produce patches of maximum
values of 255. There are relatively few values of 254.
If the least
1 The pure colors are often jarring to the eye and this is why artists often use textures
and slight imperfections to make the image more appealing. It is anyone's guess why
the optic nerve seems to react this way, but perhaps it is an effect like the moire
patterns produced when one pattern is digitized at too coarse a level. If you've ever
looked at anyone wearing a fine checked shirt or tie on television you may have seen
this effect.
Search WWH ::




Custom Search