Why watermark images?
Image
compression makes the distribution of digital images and video materials over
the Internet and on media practical. Such digital materials can b e copied and
redistributed uncontrollably. Digital image watermarking – the process of
inserting data into in image –can be used to protect the rights of their owners
in a variety of ways:
1) Copyright
identification –
provide a proof of ownership;
2) User
identification (fingerprinting)
– encode identity of legal users to encode sources of illegal copies;
3)
Authenticity determination – if the watermark will be destroyed by modification
in an image, its presence quarantines authenticity;
4)
Automated monitoring – monitor when and where images are used (for royalty collection
or the location of illegal uses);
5)
Copy protection – they can specify rules of image usage and copying Visible
watermarks
A Visible
watermark is an opaque or semi-transparent sub-image or image that is placed on
top of another image (that is watermarked) so that is obvious to the viewer.
Invisible
watermarks cannot be seen with the naked eye but they can be watermarked
image
and the recovered with an extracted
appropriate
decoding algorithm. The invisibility is assured by inserting them as visually redundant
information watermarked image after high
quality JPEG compression and the extracted watermark…
The
watermark can be recovered by zeroing 6 most significant bits of the image and
scaling the remaining values to the full intensity range.
An important
property of invisible watermarks is their resistance to attempts (both
intentional and accidental) to remove them… This type of watermarks is referred
to as fragile invisible watermarks.
If the
watermarked image is compressed and then decompressed using loss JPEG, the
watermark is destroyed. As shown in the last
Figure although the
visual information was preserved the watermark was unusable. This can be used
for image authentication.
Watermarking
techniques
Robust invisible
watermarks must
survive image modification (attacks) including image compression, linear or
non-linear filtering, cropping, resampling, rotation, printing/rescanning,
adding noise, etc.
Watermarking
techniques
Watermark
insertion and extraction can be performed in spatial domain as
shown
Previously,
or in the transform domain.
Two
watermarked images computed using DCT and Spring 2009 ELEN 4304/5365
DIP 10 p g the intensity-scaled differences between the original and
watermarked images.
Watermarking techniques Spreading watermarks across
an image’s perceptually significant frequency components, a can be made small to reduce watermark visibility. At the same
time, watermark security is high since:
1.
Watermarks are composed of pseudorandom numbers with no obvious structure,
2.
Watermarks are embedded in multiple frequency components with spatial impact
over the entire 2D image (their location is not obvious),
3.
Attacks against them tend to degrade image as well since the image’s most
important frequency components must be altered to affect the
watermarks.
ARTICLE
BY
K.NISRATH
FIRDOUSE
FINAL
YEAR IT
