Forensic Analysis Reveals Al-Qaeda's Image Doctoring

WerewolfOfVulcan writes "Wired reports that researcher Neal Krawetz revealed some very interesting things about the Al-Qaeda images broadcast in the mass media. Analysis shows that they're heavily manipulated, a discussion meant to illustrate a new technique that can spot forgery in digital media. 'Krawetz was ... able to determine that the writing on the banner behind al-Zawahiri's head was added to the image afterward. In the second picture above showing the results of the error level analysis, the light clusters on the image indicate areas of the image that were added or changed. The subtitles and logos in the upper right and lower left corners ... were all added at the same time, while the banner writing was added at a different time, likely around the same time that al-Zawahiri was added, Krawetz says.'"

Comments

msm

[stoolpigeon]

I think it has been pretty well documented in the blogging community that many of the images that the main stream media picks up and propogates are heavily altered, faked, or come from completely different events than what they claim to depict. This is not just with al-quaeda, but governments and any group that has an agenda and is media savvy - foreign or domestic.

Re:msm

[samkass]

Even before Photoshop, inappropriate use of stock footage or using visual scenes of the wrong event was pretty common and made for more exciting news. I remember in the late 80's reports of a student riot in South Korea... my father was there at the time. There may have been a few disgruntled students there at the time, but his pictures are completely different (no violence or anything.) Turns out the news companies heard about a student protest and just looked for random footage of asian students rioting and put it on the air when talking about the situation.

Of course, Final Cut Studio and Photoshop make it even easier, but the news has always been more about entertainment than information.

Re:msm

[Jimithing DMB]

This is very common. My dad worked for a power company and one of the local news organizations did a story on pollution. So after my dad talked about their scrubbers and other emissions controls (which he was very instrumental in putting in) the reporter decided that it wasn't sensationalist enough so he pulled a dirty trick. One of their power plants was right next to a steel mill so instead of the reporter doing his monologue with the power plant in the background, he and his camera man simply turned around and put the steel mill right next door in the background then proceeded to open up with "I'm here at .. generating station."

He didn't technically lie; after all he was on the property of the generating station. But the images didn't reflect the nearly nonexistant exhaust of the powerplant (a little NOx which shows up brown on certain days) but instead reflected the constant fires and smoke billowing out of the steel mill. No photoshop required.

Re:Software - Good thing.

[Hijacked Public]

Most forensic photography is digital these days and the resulting images run through verification software to prove that they are 'straight from the camera'.

The SLRs I shoot are Canons and they provide the option of "Add Original Decision Data" in their settings. Combined with Canon's data verification kit any of the images I shoot can be demonstrated to be originals, with minimal in camera image processing.

And anyone who thinks image alteration in the film world is too hard to undertake to swing a court case can't be taken seriously.

Re:Uh... "Forensic Analysis" my foot

[Hays]

Mod parent up.

Take off your tin-foil hats long enough to ask how accurate this "forensic analysis" is. Has it been independently verified? Tested with known manipulated videos? The outputs of the forensic analysis don't even look reasonable for these segments.

There has been some real (peer reviewed) research on detecting digital forgeries by Dr. Hany Farid and his lab at Dartmouth:
http://www.cs.dartmouth.edu/~farid/research/tamper ing.html

Re:Software - Good thing.

[necro81]

You might also check out the research of Hany Farid, a CS researcher at Dartmouth who has come up with algorithms to detect if an image has been doctored.