[WM]: Question on Barni Paper

["gardener" <g940305@ccit.edu.tw>]

Dear all,



I have read the paper of M. Barni, F. Bartolini, and A. Piva, = =A1=A7Improved wavelet-based watermarking through pixel-wise =
masking,=A1=A8 IEEE Transactions on Image Processing, vol. 10, no 5, pp. = 783-791, May 2001.

=20

There are some questions for me on page 785, where the authors = introduced and modified the mask construction proposed by Lewis
and = Knowles. A weighting function w is used to exploit the masking = properties of the HVS. The weighting function is half of
the = quantization step, which is the product of three terms, i.e., the = sensitivity of frequency, luminance, and texture. My
questions are:

=20

1.     Indices of Equations of (6) and (8) seem confusing to me. For =
example, if the original image with size 512*512 is decomposed into 4 = resolution levels (say, level 0 to level 3) through the
DWT, then the = size of each level 0 subband is 256*256, i.e., index range from (0,0) to = (255, 255), and the size of LL3 subband
is 32*32, i.e., index range from =
(0,0) to (31, 31). Now, if I want to check the brightness effect of = pixel (255, 255) on LL0 using equation (6), then index (32,
32) will = exceeds LL3 dimensions!?

2.     The power (0.2) of the last term of Equation (3), i.e., a measure =
of texture activity, seems different from that (0.034) of Lewis et = al=A1=A6s formula. The authors didn=A1=A6t explain the reason
why?





Best Regards,

=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D

gardener




______________________________________________________________________________

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