Re: [WM]: SS or QIM? Which is secure(Comparitively)?

[Ron van Schyndel <ronvs@cs.rmit.edu.au>]

Hi,

There are only really three ways that one inserts a watermark.
1. by selective replacement
2. by selective addition (which may include a transformed component of the host) 3. by selective quantisation either to the host, or
some reversibly transformed version of the host.

1. (bit-masking, some visible watermarks) This is generally accepted as being non-useful (as everyone who (mis-)quotes my 1993-4
paper seems eager to point out :( - I also advocated bit-addition).  These usually fall into the category of replacing bit-planes.
More advanced forms will insert visible watermarks (like the IBM vatican project), or even hide real content within an image.

2. (SS) This originally came from 'unjammable' radio research in the 1940-50's (with a prime patent held by a Hollywood actress.).
Its primary feature is its noise immunity.  An SS signal is best jammed by another SS signal.

3. (QIM) The central idea of QIM is not really that complex.  In possibly its simplest form, QIM might say "round all pixels to even
numbers, then add 1 for the white pixels in the binary watermark image - assuming same image dimensions for host and watermark".  

Hence 
  I_w(x,y) = AND(I_host(x,y),1)    for all x,y
reveals the watermark image (after scaling).  


====

Clearly all methods modify the image.  And all do so in different ways.  The security depends heavily on what kind of attacks are
expected and how much and what kind of damage is tolerable.

Ultimately, in the majority of cases, the question of watermark attack resilience (and the digital piracy that it is used against)
is one of economics, more than security.





(PS.  If applied as a stream, #3 is actually FM modulation.  Really! )

Bhanu Prasad wrote:
> Hi,
>     I am working on my Masters Thesis.It is based on SS audio 
> watermarking.SS and QIM seem to be the dominating technologies.SS is simple.QIM seems to be complex.
> 
> Which one has the future?

-- 

Cheers, Ron

============================

Ron van Schyndel,  Lecturer,
Room 14.6.4, School of Comp Sci & IT, RMIT University GPO Box 2476v Melbourne, Victoria 3001, Australia ronvs@cs.rmit.edu:.au
http://www.cs.rmit.edu.au/~ronvs/


______________________________________________________________________________

Watermarking Mailing List - http://www.watermarkingworld.org/ml.html
To unsubscribe send email to "majordomo@watermarkingworld.org" with
"unsubscribe watermarking YOURMAIL" in the body.
______________________________________________________________________________