|
|
|
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE[2]: [WM]: region based watermarking question-SOME IDEAS
The paper does not provide much detail about the experiment. So I guess the best person to give you a precise answer will be the
authors.
Anyway a possible explanation can be as follows. In each experiment they embed a watermark, say 'WMn' and try to detect all
watermarks WM1. WM99 (and many more random watermarks not included in the samples, which is not uncommon). This gives a FAR and FRR
FOR THAT PARTICULAR WATERMARK. Repeating this for all 100 sample watermarks gives 100 FAR and FRR for 100 watermarks.
The final values for FAR and FRR will be a function of these values (e.g.
average of them).
Kourosh
-----Original Message-----
From: SIAMAK YOUSEFI [mailto:syousefy@yahoo.com]
Sent: Tuesday, 21 June 2005 8:19 PM
To: Kourosh Fallah
Subject: RE: [WM]: region based watermarking question-SOME IDEAS
Dear Kourosh
Thanks for replying
As you said, there are only 100 watermarks. so with
your assuumption, if one error out of 100 traial
occured, the FAR= 0.01 and for other cases above this
number, but you see numbers such as 2.6564*10^-4.
How you analyze these numbers?
by the way, I think they map the numbers to the normal
distribution variable and then evaluate the
probability.
I almost forgot to ask you, Where are you?Iran or not?
best regards,
yousefi
--- Kourosh Fallah <kourosh@ics.mq.edu.au> wrote:
>
>
> Dear Siamak,
>
>
>
> On the title of table 1 in "robust watermarking on
> Facial Images...." paper,
> it says (COMMON DETECTION THERSHOLD=0.32). Assuming
> they have used the same
> threshold for all the images, finding FAR and FRR
> could be just dividing the
> number of False Acceptance (detecting watermark when
> there is none) and
> False Rejection(not detecting any watermark when
> there is one) cases to the
> total number of watermarked images.
>
>
>
> You might like to use contact information given for
> I. Pitas here
> <http://poseidon.csd.auth.gr/EN/> (People>Faculty).
>
>
>
> Regards,
>
> Kourosh
>
>
>
>
>
> -----Original Message-----
> From: watermarking-owner@watermarkingworld.org
> [mailto:watermarking-owner@watermarkingworld.org] On
> Behalf Of SIAMAK
> YOUSEFI
> Sent: Sunday, 19 June 2005 9:18 PM
> To: watermarking@watermarkingworld.org
> Subject: [WM]: region based watermarking question
>
>
>
> Dear members
>
>
>
> I am MS student of electrical engineering.
>
> I have worked on digital image watermarking more
> than a year. I started by
> pitas papers. By the way, I read papers, "robust
>
> watermarking of facial images based on salient
> geometric pattern matching"
> and "region based image watermarking" by pitas and
>
> nikolaidis several times.
>
> but I couldn't underestand, how they compute the
> results of tables(FRR and
> FAR). in other way they embedd 100 different
> watermarks
>
> in an image and then observe the detector output.
> after that they reach a
> table of FARs and FRRs that I couldn't underestand
> how
>
> they obtain it.
>
> I didn't find the newest email of nikolaidis.
>
> any help in finding new email or helping me about
> the problem is
> appreciated.
>
>
>
> best regards,
>
> yousefi
______________________________________________________________________________
Watermarking Mailing List - http://www.watermarkingworld.org/ml.html
To unsubscribe send email to "majordomo@watermarkingworld.org" with
"unsubscribe watermarking YOURMAIL" in the body.
______________________________________________________________________________
|
