[WM]: Re: Reply to Andrea de Polo & Ismail Ahmad's questions about "sw to compare image quality"

["Zhou Wang" <zhouwang@ieee.org>]

Hi Vural,
Thanks for your interest. I'm trying to answer your questions:

> I saw the homepage comparing image quality.
> However I see one picture for jpeg/jpeg2000. Do you have results of 
> jpeg
and
> jpeg2000?

I gave 2 sample images in the page, one JPEG, and one JPEG2000. In the JPEG/JPEG2000 scatter plots, the '*"s represent JPEG
images, and the "+"s represent JPEG2000 images. For quantitative evaluation about the performance such as correlation
coefficients, please look at the paper http://www.cns.nyu.edu/~zwang/files/papers/ssim.html


> As far as I see the still images, there is a huge difference between
original image
> and jpeg compression. There should be so small difference that you 
> can't distinguish.

I show these highly compressed images for visibility purpose. In the paper, you see more examples at different quality levels.
There are a lot of high quality images in the database.

In fact, the images I show in the page is not as bad as you see because they are cropped and enlarged from high resolution images
(for visibility of course). The images in their original sizes are actually not so unacceptable as you see, especially for, say
video over the Internet.


> By the way, there is matlab source but do you have C source for 
> those,too?

I was told that somebody implemented the algorithm using C and made a nice tool for the users to analysis their images, for both
still images and video sequences, and they put the code for free on the web somewhere. But I can't find it myself! Maybe the DIVX
video coding people did it? Hope somebody in this mailing list can help.


> Thank you
> SADI
>






______________________________________________________________________________

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