Re: [WM]: Watson Model Improvement

Dear Mehul,

Sorry for the delay. Mybe you're right in whole-image processing. But I'm affraid if there is a well-known model for contrast masking and luminance masking in the literature. So you should be cautious in evalutaing JND values, by performing meticuluos experiments. also by changing the size of the image, JND values should be changed. While in block processing, JND's are deterministic for the blocks.

In the paper I mentioned previously, a third type masking efect has been considered to evaluate the watermark robustness in the IA-DCT scheme. This is attached to this email.

I'd be happy to be informed about your research results.

Good luck and Regards,

Amir

Multimedia Lab.

EE Department

Sharif Univ of Tech.

Tehran, Iran

ul@scet.ac.in wrote:

Dear Amir,
Good to get the response from you. But my experimentation results are
based on using JPEG qunatization matrix instead of standard table takne by
Watson and there is improvement in the performance. I am also computing
both luminance and contrast masking. May be you are correct in pointing to
block processing but some experimentation suggest that block processing
leads to artifacts whicj i am avoiding. I do not have access to paper you
mentioned can you send it to me.

Regards and Thanks for response

Mehul
> Hi,
> if u use only JPEG quantization metrix, you ignore luminance masking and
> contrast masking, but just using frequency sensitivity. Also, you cannt
> perform addaptive watermarking on the whole image, block-wise processing
> leads to addaption of the mark to the host image.
> you can also ! see this as a trend in improving adaptive watermarking:
>
> HOUMANSADR, A., and GHAEMMAGHAMI, S., "Robustness enhancement of content
> based watermarks using entropy masking effect," Lecture Notes in Computer
> Science 3710, pp. 444-458
>
> Regards,
> Amir
>
> mehul@scet.ac.in wrote:
> Hi,
> I have been recently working on improving the watson model for adaptive
> watermarking. I am utilising standard JPEG quantization
> matrix instead of intensity values suggested by the watson. If I would be
> processing whole image instead of block processing as
> suggested by Watson method Do I expect some improvement in timing of
> execution.Also Do i expect better performance of scheme against
> JPEG compression. If so why?.
>
> Thanking you
>
>
> Mehul S Raval
> Department of E&C,
> Sarvajanik College of Engg and Tech,
> Surat
>
>
>
>
> ______________________________________________________________________________
>
> Watermarking Mailing List - http://www.watermarkingworld.org/ml.html
> To unsubscribe send email to "majordomo@watermarkingworld.org" with
> "unsubscribe watermarking YOURMAIL" in the body.
> ______________________________________________________________________________
>
> Send instant messages to your online friends http://uk.messenger.yahoo.com