[WM]: Questions on a paper in image hashing

[YU <longjiang_yu@yahoo.com.cn>]

Dear Watermarkers,

Robust perceptual image hashing is relevant to image watermarking. It may be useful in watermark detection or decoding. 

I'm a Chinese student from Harbin Institute of Technology. I'm working on image hashing. The following paper,

V. Monga, D. Vats, and B. L. Evans, "Image Authentication Under Geometric Attacks Via Structure Matching", Proc. IEEE Int.Conf. on
Multimedia & Expo, Jul. 6-8, 2005, Amsterdam, The Netherlands, accepted for publication, available at:
www.ece.utexas.edu/~bevans/papers/2005/authentication 

claims signature extracted is resistant to geometric attacks. They use 256x256 Lena, Bridge and Peppers as test images. And Bridge
is rotated by 25 degree to show robustness (the other two are rotated by 15 degree). MATLAB code for the authentication scheme
described in this paper is available at:
http://www.ece.utexas.edu/~bevans/projects/hashing/software.html

I've downloaded this code and found it robust when Bridge rotated by 25 degree. But images with different size such as 512x512 Lena
are not robust. Even when Bridge rotated by 5 degree, it doesn't work. Lena and Peppers don't work either when rotated by 5 degree.

I'm wondering why. Could some one explain for that? 

Thank you very much!

Best Regards,
Longjiang Yu


______________________________________________________________________________

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