Detecting Faked Photographs Gets Easier

nusratt writes "Some years ago, an issue of 'Whole Earth' had a convincing cover-photo of a flying saucer cruising low over downtown San Francisco in broad daylight. The accompanying feature article proclaimed that photographs can no longer be trusted as evidence of anything, because of the ease of doctoring images digitally and undetectably. Now, Dartmouth Professor Hany Farid and graduate student Alin Popescu 'have developed a mathematical technique to tell the difference between a "real" image and one that's been fiddled with.' Farid says, 'as more authentication tools are developed it will become increasingly more difficult to create convincing digital forgeries'." There's also an NYT story.

Comments

Seamless Math Next?

[mfh]

While it may become increasingly difficult to forge digital images, and even forge hard currency, the result could be of two possibilities;

1. Forgers get smart and use older cameras to take a picture of a digital forgery to pass as an original, using blurring techniques offered by physical means and lens... etc (easy)
2. Forgers quit being forgers (unlikely)
3. Alteration technologists create armor against image forgery detection algorithms (possible)

For me, I think any time spent trying to beat the detection of forgeries would be a good thing in terms of art and creativity -- not to mention the possibility of better digital growth algorithms to join layers mathematically seamlessly (which could be used in games and simulation engines for better realism). However, law enforcement agencies might try to combat the circumvention of forgery detection by charging people with crimes for only trying to make their images more realistic and improve technology. It's a messy issue, that will sort itself out over time.

In Doom 3 Bloopers, a mod I've started on, I am looking at ways of integrating realworld imagery into the mod, and this detection stuff could actually help me to better integrate my own art and images if I can find a way around it. Let's face it, if the math says it's an original, the human eye will be fooled, which is the goal of most video game design. If anyone wants to help along those lines, they should contact me!

Re:Seamless Math Next?

[Anonymous Coward]

I gotta say, while the first half of your comment was certainly insightful, that was one of the most blatant attempts to force your pet project into an unrelated comment I've ever seen!

You, sir, have balls!

Re:Seamless Math Next?

[Ieshan]

It seems as though the algorithm they have works on the fact that there's some statistical difference between a real image and a faked one, but not 100% of the time.

Why wouldn't someone simply build a filter into a program that changed bits in an image until it passed a check by their algorithm - on failing, it would simply go back and change more appropriate bits?

Seems as though it would be a computationally intensive but a logically easy task.

Re:Seamless Math Next?

[yintercept]

My first thought on the article was the same. The mathematical tests to determine fake photos will have value only until the fake photo industry builds the tests into their software. For that matter, I suspect that the main use of the Hany Falid method will be to make your fake photos even more realistic.

The most interesting line in the article was:

but computers make it easier for more and more people to manipulate images.

One could read into these lines that the ability to fake photographs was great until anyone could do it. Now that we know how easy it is to fake photographs, we no longer implicitly trust messages...but we will trust mathematically authenticated fake photographs because math is infallable.

Re:Seamless Math Next?

[freshmkr]

Let's face it, if the math says it's an original, the human eye will be fooled, which is the goal of most video game design.

Don't be so sure about that!

I don't know what these researchers are doing, but I can venture a guess. You can think of an image as being composed of a large set of superimposed sine waves---they look like this. To figure out what sine waves make up an image, you do a Fourier Transform, which is well documented on Google.

Natural images contain a characteristic power spectrum: some frequencies--lower ones--tend to occur quite often, while others--higher frequencies--are less common. The spectrum is actually pretty regular across an image set. I'm betting, though, that fake images don't respect this power spectrum and lead to detectable anomalies.

But beware! You can have completely bogus images that also respect the power spectrum. Some researchers at MIT (Torralba et al.) use power spectra to successfully detect different image environments, e.g. indoors, outside in a city, out in the country, etc., but in the papers they show some images that have been reconstructed from their spectral models.

You would not be fooled. They look like finger paint pictures.

--Tom

Re:Seamless Math Next?

[B'Trey]

One could read into these lines that the ability to fake photographs was great until anyone could do it. Now that we know how easy it is to fake photographs, we no longer implicitly trust messages...but we will trust mathematically authenticated fake photographs because math is infallable.

Or you could read into them that when it was rare and difficult to fake photographys, most of the photographs you saw were genuine, so you could place a decent amount of trust in what you were seeing. Now that faking photos is easy and commonplace, you can no longer place much trust in photos. With mathematically verified photos, you can place more (though not complete) trust back in the photo. It isn't foolproof but the level of assurance is significantly higher.

Re:Seamless Math Next?

[Megahurts]

One could read into these lines that the ability to fake photographs was great until anyone could do it. Now that we know how easy it is to fake photographs, we no longer implicitly trust messages...but we will trust mathematically authenticated fake photographs because math is infallable.

Or you could read into them that when it was rare and difficult to fake photographys, most of the photographs you saw were genuine, so you could place a decent amount of trust in what you were seeing. Now that faking photos is easy and commonplace, you can no longer place much trust in photos. With mathematically verified photos, you can place more (though not complete) trust back in the photo. It isn't foolproof but the level of assurance is significantly higher.

Or you could realize there's inherent error in any statistical method, and that with a little bitof foresight, this error could be expoited to provide false results. The best detector of falsified photography is still a well-trained human eye. The key to knowing whether or not to trust images is to train your eye.

Re:Seamless Math Next?

[NanoGator]

"Seems as though it would be a computationally intensive but a logically easy task."

Probably. Consider this, though: Fingerprints are common knowledge. We ALL know that if we commit a crime, fingerprints will be lifted to try to catch us. There are well known ways to defeat this, but remarkably, LOTS of people are still leaving fingerprints as clues at a crime.

I hope you can forgive me for reading a little more into your post than you actually said. I don't know for sure if you were going the "this could easily be defeated, thus it is ineffective" route. But I thought this would be as good of time as any to mention this.

Re:Seamless Math Next?

[yintercept]

when it was rare and difficult to fake photographys

It was never rare nor difficult to "fake photographs." From the very moment photography existed, people realized that they could control the message the photograph carried. Most of the early photographers had studied optics and developed their own film and manipulated images accordingly.

Most of the early photos you come across were staged. Taking a photograph was a big deal. People dressed up in their best outfits and the photographer would construct the scene. Often the photographer colored or otherwise enhanced the image. Photography has always been an art form. As an art form, people choose the message to convey.

Using photographs as evidence has always been problematic. The struggle is for lawyers to keep enough faith in photos to be able to use them in court. Personally, I think having a method to declare a photography mathematically correct immediately creates a problem where people with sufficient resources to fake mathematically correct fake photos will have the ability to manipulate the courts.

There never was that much faith in photographic evidence. I think we are better of having doubts about photographic evidence than we will be if we sanctify any photos as mathematically correct. The methods will have some value in quickly identifying tampered evidence, but will not have value in verifying it.

Re:Seamless Math Next?

[B'Trey]

OK, if we're going to have this conversation, we need to define what we mean by "fake photograph." Yes, many photos were staged, particularly when taking photos was significantly more difficult than "point and shoot." But a staged photograph is not a fake photograph - it's still a reasonably accurate representation of reality as it existed at the time the photo was taken.

You're also quite correct that photos are routinely manipulated in the dark room. However, manipulating the color and otherwise enhancing the image is not at all what most people mean by a "fake photograph." There's a fundamental difference in those types of manipulations and putting Sarah Michelle Gellar's head on a porn star's body or putting John Kerry and Jane Fonda on the same podium together. This type of thing was possible before, but it was much, much more difficult and much less common.

Re:Seamless Math Next?

[angst_ridden_hipster]

At the Getty Museum, they recently had an exhibit of influential photographers and the history of photography (http://www.getty.edu/art/exhibitions/genius/).

One example photo from the beginning of the last century or before, depicted two figures by a lamppost on a foggy Parisian street. I took it for an unaltered print, but it turned out to be a composite of seven [!] separate images. It was laboriously done in the darkroom. It was incredible. (Unfortunately, I can't find the specific photo on their otherwise excellent site.)

Now, we've all seen great darkroom manipulation like the work of Jerry Uelsman (www.uelsman.net), but this particular picture was a hundred years older than his work. It was absolutely convincing.

I guess my point is, photographs have in fact been "faked" as long as there have been photographs.

Actually...

[julesh]

I think authentication tools make it easier. As someone who's tried a little photo manipulation in the past, I can tell you that the hardest thing is knowing when something's right. If you have an automated tool that can tell you when it's right, it becomes easier. Of course, that relies on the tools working...

Self Defeating

[Dozix007]

While an equation to make sure that a photo is not forged is all well and good, it is self deafeating. Just like simple encryption, which is good for people with simple problems, exploring the equation will yield a way to fool it. If you have someone who truely wants to forge a photo, they probably could still defeat the check.

Re:Self Defeating

[ruszka]

Which makes me wonder.. did they create this in order to detect forgeries? or possibly to make it easier for the government to forge photographs and then use the technology to make those images "authentic"

Re:Self Defeating

[slavemowgli]

I know how cryptographic (and hashing) algorithms work.

As for "almost unbreakable" - well, that simply means that it *is* breakable after all; brute force does work, even though it's usually inefficient (which is why I added that an attacker would need a sufficient amount of time, energy and/or computing power).

Furthermore, from what I know, it is neither proven that your average hash functions (MD5, SHA1, RIPEMD-160 etc.) are collision-free nor that the underlying mathematical problems of your average asymmetric ciphers (RSA, DS, El Gamal etc.) indeed cannot be solved with less effort than brute force. (I may be wrong here, so feel free to correct me)

And lastly... you missed my point completely. :) What I really was saying was that the fact that there will be ways to defeat this system does not mean it is completely useless or obsolete. Without looking at the algorithm in detail, it's not even possible to say that there likely will be easy ways around it (common sense might dictate there are, but common sense also dictates that RSA is easily breakable - mine does, anyway). Ultimately, only time will tell just how feasible it is to defeat this system, and only that will determine just how useful it will prove to be on a given level. (As an example, I could very well imagine that an intelligence agency or a similar organization will be able to produce images that pass the test when necessary, for example, while your average Paintshop Pro user won't).

So much for my faked photo of Rumsfeld & Sadda

[The I Shing]

Now they'll be able to prove that my photo of Rumsfeld shaking hands with Saddam in 1983 is a fake.

Rumsfeld was really shaking hands with an alien, and Saddam was shaking hands with Elvis, but the resulting merger of the two photos was much more provocative.

Just a thought...

[julesh]

Farid and his students have built a statistical model that captures the mathematical regularities inherent in natural images.

I wonder if they've considered the potential applications in image compression?

I had developed a foolproof algorithm...

[Anonymous Coward]

...but it was only accurate when the photo contained an image of Admiral Ackbar.

What kind of digitized photos does this work on?

[multiplexo]

If I take a high resolution TIFF image and start mucking with it I can see how it would be easy to find the manipulations (especially if you're as thumb-fingered as I am with PhotoShop). However with most digital cameras using various compression schemes to store the images how can you tell what is a result of manipulation versus what is an artifact of compression and or digitization? Certainly some gross manipulations will be obvious compared to the properties of compressed images, but I would imagine at some point you'd be hard pressed to say "this image was deliberately manipulated" instead of "this is a compression or digitization artifact".

While this might not be a problem for gross manipulations (the faked John Kerry/Jane Fonda photo being a recent example) I can imagine a class of images where subtle manipulations caused great effects and were not readily distinguishable from compression artifacts.

Okay, but here's my question.

[Anonymous Coward]

Farid's algorithm looks for the evidence inevitably left behind after image tinkering. Statistical clues lurk in all digital images, and the ones that have been tampered with contain altered statistics.

That makes sense. However, it seems like these statistics would be based on very minor details. What happens if you run their analyzer on an image that has been altered by using lossy image compression, such as JPEG compression? Lossy image compression is designed to obliterate details humans wouldn't notice; some of these details might be significant to their statistical model. Would JPEG-compressing an image make it impossible to determine its veracity? Would their software just tag all JPEG images with compression below a certain threshold as being "unnatural" (since they have been, after all, digitally altered-- just not digitally altered in a content-relevant way...)

And don't some digital cameras use lossy formats such as JPEG as their native storage format?

The unfortunate thing

[Anonymous Coward]

The article says the research is founded by the Department of Homeland Security. That means that despite the many useful possible aspects of this technology, it will probably never see the light of day. If the DHS is going to be using this technology to identify faked photos, it would be greatly in their interest for the full algorithm and its implementations to not be made available to the public-- since after all if people know what the DHS is using to determine faked photos, they can target the algorithm.

Now that I have written that, looking around, it appears I am actually wrong. If you look at Mr. Farid's personal page, it appears he will be publically presenting a paper covering the fakeness-detection algorithm. I hope the full algorithm will be presented to the academic community.

Why not *make* it real?

[over_exposed]

So if it detects digitally altered images, why can't we just 1) Alter the image to our heart's content 2) Print it on a high quality printer ($500- $2K at CDW) 3) Scan it back in with a nice scanner

A more in-depth article

[asterism]

If you are looking for more detailed information, along with equasions, here is a link to one of their recent publications on the topic: http://www.cs.dartmouth.edu/~farid/publications/sp 04.html

practical use

[Slothy]

Run it on the goatse guy! If the results are positive I can finally start sleeping again.

Admissibility of Digital photographs as evidence?

[walmass]

There are stories about successful defense against digital photographs in criminal cases. "Enhancements" using photoshop can be considered evidence tampering. So this technique can have a life-altering implication for some people.

Re:What kind of digitized photos does this work on

[blonde rser]

The NY Times article deals with this directly.

But Professor Farid said that for now the technique does not work as well with files created in JPEG, the compressed picture format most commonly used online. As the size of a JPEG file shrinks, the correlations between pixels become much less obvious. "At 90 percent quality, it falls apart very quickly," Professor Farid noted.

So good point. That does seem to be a problem. The NY Times article has more details than the other; it is worth reading.

Re:MICHAEL, thanks for adding the . . .

[julesh]

No, it couldn't. All it can prove is that if an image is fake it has been done very well.

If A doesn't conform to the statistical distribution of B, then A isn't B (with a high degree of confidence). But if does, that doesn't mean it is B -- you might just be looking at the wrong set of identifying features. I.e. not everything with two feet and a bill is an aquatic bird; it might be the waiter.

from the author

[Hany Farid]

In response to some of the posts on our work, we have developed a number of different techniques for detecting certain types of tampering. For those interested in technical details, please see:

sp04.html
ih04.html
sacv03.html

And, we have two new papers currently in review (abstracts are currently on-line, and preprints will be available soon):

sp05a.html
sp05b.html

Some of these techniques work, as some have pointed out, only on high-quality jpeg or uncompressed images, while others work on lower-quality images. We are only in the early stages of development, and are currently working to extend some of these ideas to low-quality jpeg and gif images (though this will likely be a harder problem given that the compression artifacts will overwhelm any statistical perturbation resulting from tampering). One outcome of this may be that a legal standard is set that enforces images brought into a court of law to be of a certain resolution and compression quality.

I will be the first to admit that each of the techniques that we have developed can be reverse-engineered, though doing so is more difficult for some techniques than others. It is our hope, however, that as we and others continue to develop more techniques it will become increasingly more difficult (though never impossible) to simultaneously foil each of the detection tools.