[WM]: confusing about N(0,1)random sequence=20

[nobel_meg <nobel_meg@sina.com>]

Hi, everybody ,i am a greenhand in digital watermarking and now i am =
confused about the N(0,1)random sequence .in some papers,
they said use the distribution of N(0,1),with the length of 1000,and =
then embeded into the original image ,X'(u,v)=3DX(u,v)(1+a.W)
where X'and X are coefficents of the =A3=C4=A3=C3=A3=D4domain say in DCT =
domain and then use the eqution as w'=3D(X'-X)/a*X. then calculate the
correlation as sim=3D(W*W')/sqrt(W*w'). i do the experiment of 1000 of =
the length of 1000 N(0,1),and among them ,two are the same
and calculate the Sim ,but it only up to 30, it shouldn't come to 50 =
..and it was the same sequence .so if the recovered water and
the orignal water, to do to the sim, it couldnt reach 50. my another =
question is that the watermark recovered from the coeffients
of the DCT are not the sequence of the original ,not roughly the same =
,some value are large and some are small. for example ,if i
embed 1 into the original iamge , and i could get the W as 88.99 or =
other posive value. if i embed 0, i get 0.but if i embed -1, i
would get a value that is smaller than 0. so my question is that whether =
we should to do some qualiztion about the recovered
watermark to get a better performace ?=20
if someone had the Matlab souce code of the paper related with DCT or =
DWT  and use the above mentioned algorithm, pleease help me=20

sincerely yours=20
                                  margaret =
______________________________________




______________________________________________________________________________

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