2. Personal Greenspun set: The Personal Greenspun images can be
downloaded from http://www.photo.net/philip-greenspun/
photos/digiphotos/. The thumbnails as well as the filename
of the 400 Personal Greenspun images can be found in the Per-
sonal Greenspun thumbnail page.
² Google set: The Google images can be downloaded by following the
hyperlinks or the URLs listed in the Google thumbnails page, accessible
from the dataset website.
² Recaptured PRCG set: As the recaptured PRCG are the reproduc-
tion of the PRCG, the distribution of these images is limited by the
copyright. As a result, the Recaptured PRCG will not be distributed.
9 Conclusions
This technical report explains the needs for a new dataset catering for the
PIM versus PRCG classification problem of the passive-blind image authen-
tication research. During the process of working on the mentioned problem,
we have designed and implemented a dataset that specifically takes into
account the unbiased diversity of content, reliability of sources, and ad-
vanced CG rendering techniques. This Columbia Photographic Images and
Photorealistic Computer Graphics Dataset is to be made available to the
research community and this technical report will serve as a user guide for
the dataset. We make the data set open and publicly accessible in order to
promote active interest and broad collaboration in this exciting area.
10 Acknowledgements
This project is supported in part by NSF CyberTrust program (IIS-04-
30258) and Singapore A*STAR scholarship (for the first author). The au-
thors thank Philip Greenspun for granting us the permission of using his
personal images and Lexing Xie for her kind help.
References
[1] I. J. Cox, M. L. Miller, and J. A. Bloom, Digital Watermarking, ser.
The Morgan Kaufmann Series in Multimedia Information and Systems.
Morgan Kaufmann, 2002.
21