[WM]: How to calculate the HVS? Should normalize the DWT coefficient first?

[zh_do@encs.concordia.ca]

Dear all,

After decomposing the host image into DWT sub bands, I want to find the most significant coefficient in the detailed sub band.

There is a method using HVS. We can caculate the luminance, the frequency and oritation, the apporimaxi to the edge and texture,and
multiply them together to get the quantizer step, the half of the quantizer step is the weight factor. We then add the emmbeded logo
image into the significant coefficient, with the weight factor. All these can be seen "Improved Wavelet-Based Watermarking through
pixel-wise masking" by Maruro Barni.

But my question is: after the decomposition of the host image, shall we normalize the sub bands, from 0 to 255? Because some values
in the subband are very large, more than 255, while others are too small and negative. We can not use the sub band to calculate the
HVS factor directly!!

If I only normalize the HH subband, how about the detailed subband? If I normalize all the subbands to get the weight factors, these
weight factors are the normalized sub band factors. Then in the embbeding step, I add the logo bits to the normalized subbands? And
denormalizing the embedded sum of the host subband and logo image, inverse DWT to get the final image?

Could anybody tell me?

Thank you very much!





______________________________________________________________________________

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