Browse Prior Art Database

Encoded Security Feature for Digital Images

IP.com Disclosure Number: IPCOM000117945D
Original Publication Date: 1996-Jul-01
Included in the Prior Art Database: 2005-Mar-31
Document File: 4 page(s) / 145K

Publishing Venue

IBM

Related People

Beeteson, J: AUTHOR [+2]

Abstract

Cameras and photographic images are increasingly using a digital representation of the data forming the image, rather than the traditional chemical methods employed in film. The images produced by, for example, security cameras are admissible evidence in court which may lead to a conviction.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 41% of the total text.

Encoded Security Feature for Digital Images

      Cameras and photographic images are increasingly using a
digital representation of the data forming the image, rather than the
traditional chemical methods employed in film.  The images produced
by, for example, security cameras are admissible evidence in court
which may lead to a conviction.

      Various video authoring tools are now readily available for
the PC at relatively low cost and can be used to modify the digital
data representing the image.  Indeed, so advanced are some software
packages that they may be used to completely alter the original image
e.g., by substituting one persons face from one image to another.
Once the work on the image is complete there is no way of knowing
whether this is an original or is a creation.

      Thus, undetectable malicious or fraudulent behavior is
possible.

      To overcome the problem described above, a means of ensuring
authenticity of the original image is disclosed.  There are three
keys steps in this process:
  1.  The digital data representing the image is applied to an
       algorithm which will create a unique "signature" indicative of
       the data content within the image.
  2.  This signature is then scrambled using a known cipher.
  3.  The encrypted data is then embedded within the image data in a
       manner and position determined by the cipher key.

To elaborate on each of these steps now:

      By considering each pixel of the image as an n bit word whose
data represents the colour information relating to said pixel, a bit
stream may be formed which is passed into, for example, a cyclic
redundancy checker.  Concatenating sequential pixels forms a bit
stream of the whole image or part thereof.  Once the bit stream has
been passed through the CRC hardware, a unique code will be left on
the output of the CRC hardware (or software algorithm) representative
of and unique to the image data.

      In the simplest form of the concept, this information could be
stored in binary form by setting the least significant bits of
successive pixels in the image data to correspond to the CRC.  There
are many methods that could be employed, line by line, column by
column, a diagonal line covering both lines and columns or whatever.
It should be noted that in "real world" images (as opposed to those
generated by computer), seldom do 2 pixels have identical values due
to, for instance, variations in ambient lighting, shadows and the
like.  Thus, a change in the least significant bit by the storage of
the CRC will be invisible to the viewer.  This is especially so when
considering that PhotoCD type images use a minimum of 24 bits for
each pixel and that primary shades only are rarely, if ever,
encountered.

      It will be appreciated that a problem with the method described
above is that it is now impossible to restore the original image data
such that the original CRC may be recr...