[STOCKPHOTO] Friendly debate

[Russell Graves]

Hello,

I am debating a person as to the authenticity of a jpeg photo that
was sent to me via e-mail. I am no imaging pro and I was wondering
if anyone knows whether or not you can tell a picture has been
doctored by looking at it with a software package.

If so, does anyone care to take a look at it and tell me whether it
is authentic.

Thanks,

Russell Graves
www.russellgraves.com

------------------------------------------------------------------------
Sponsored by The STOCKPHOTO Network - http://www.stockphoto.net/
------------------------------------------------------------------------

[Jeff Boucher]

Sure...I'll bite - where is it?
Cheers
Jeff Boucher

[Daryl G. Jurbala]

One can usually tell, but it depends on the image, resolution, and what
doctoring may have been done.

Fell free to send it to me off-list. If it's pretty apparent I can tell
within a minute or two.
Daryl

[Kevin Breckenridge]

There are many ways to detect manipulation,

Considering jpeg and every other compression algorithm introduces
artifacts I rely on, pixel x pixel & mass pixel based anomalies based
on visual and numerical characteristics,

i.e.: - mass blurring, inconsistent sharpness, noise, contrast,
contour alignment, uniform color description, like LAB luminosity values
and/or color space plots.

Sometimes the presence of undeleted paths & alpha channels are the
clincher! Then again sometimes if just looks plain goofy...it is!

KB,

[Gary Gaugler]

Two relatively easy ways to analyze a digital image is a
forward FFT and computing the Point Spread Function.
Sometimes just one will do the job. The way I do it
(not all that often) is to look for non-monotonicity.
There are some tweaking adjustments necessary at times,
like changing next-neighbor values, but overall, it seems
to work.

gary g.