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?

[perlchild]

I think point 3 will come true first, as the truly "skilled" forgers will come to notice a lot of those "real world" factors as light correction (joining two pictures with differing angles to light sources). Layer-fudging is certainly interesting, when it comes to overlap, but I'd be surprised if light and reflection don't also play a role in the algorithm.

Perhaps some of the forgers(maybe just the ones who used to forge paintings *tongue in cheek humor*) already do pay such attention, after all, attention to detail certainly is a skill they would do well to have, and more than a few tv programs/movies mention how a reflection that looked "wrong" helped a policeman catch a criminal attempting a "perfect" crime.

However, and to recenter my argument on something closer to the article, for authentiying currency, which normally does not use "real-world" photos(an hologram is not a real world photo, as it uses artificial laser light instead of normal lighting), this technique may be of limited help as-is.

However, this research should help those researchers looking at making currency "forgery-proof" consider other alternatives, using the same algorithm. Imagine a real world picture, that follows the equations, but that uses an encryption key for input, perhaps a MD5 hash of the serial number on the bill. You'd have a picture that'd look weird of course, but without knowing the equations, the key, and having a reasonable fac-simile of the original, it would make serial-number "fudging" a lot harder. Of course, perhaps those currency printers(the independant ones, who work for/with countries to make currencies harder to forge) are already hating me right now, for even suggesting money printing costs per-bill should go up.

From this point of view, the algorithm would certainly be implementable, in read-only fashion in business-card-scanner size devices. That might help improve confidence in currency safety(I'm trying to wrap my head around a way to improve credit and debit cards, but they all involve a lot more steps than what we have right now).

One silly idea: print a digital photo of a credit card on a credit card, scan the card, see if it matches the algorithm, and how well the algorithm detects alteration. Thinking of a recursive picture printed on credit cards makes my head hurt though.

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?

[bkhl]

Let's face it, if the math says it's an original, the human eye will be fooled

No. These formulas doesn't, obviously, consider the contents of the images, only the distribution of pixels (or whatever) within them.

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?

[Anonymous Coward]

It seems to me the obvious next step is to take these supposed
forsenic software suites, find out what they are doing, and use their own algorithms to make a filter to change the giveaway
flaws so they don't tip off the software the item has charcateristsics that spell 'fake'. Then we get into a sort of virus/anti-virus war. The bad guys diddle their filters, the good guys have to create new algorithms.

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?

[Pieroxy]

One of the best algorithm I have found is to compress the image in a low-quality JPEG. The JPEG scheme of compression will detect stuff that are just invisible (or barely visible) with a human eye. Most of the time, the artifacts will choke on the part of the picture that was artificially touched, making them extremely visible.

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?

[Anonymous Coward]

OK, here's a good example: It was so difficult that a couple of girls, aged 16 and 9, pulled it off back in 1917. See http://www.unexplained-mysteries.com/articlecottin gly.shtml

Re:Seamless Math Next?

[russotto]

I think what you're looking at there is an artifact of the originals of the doctored images being JPEG-compressed in the first place. Modifying the image creates discontinuities in the JPEG artifacts, which re-compression will tend to make very evident.

An image well-faked from lossless originals would not exhibit that behavior.

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.

Re:Seamless Math Next?

[jrockway]

What? I think it made perfect sense. If he's not lying... But it sounds feasable and interesting to me.

Re:Seamless Math Next?

[yintercept]

I agree that people are having more fun with their digital cameras than ever before.

In discussing fake photos we need to separate the intent to deceive from simple adjustments or artistic statements.

My point is that in order for a person to effectively deceive others, they must first have a way to claim legitimacy for a photo. Creating a program that "mathematically" verifies a photo opens up the door to deceit. You can't argue with a mathematically verified photo.

Now then, if we were to get into a question of what percent of photos were "faked" my guess is that a larger percent of photos taken in the first years of photography were faked than are today. At the height of the Kodak Instamatic years, there was probably a low in the number of faked photos. The reason for my assertions is that the first photographers really had to know their art to take pictures. Today, however, people are taking billions of photos a year. Before I dropped my camera, I had taken over 7500 pictures with it.

I concede that the number of pictures faked for the purpose of manipulating others has risen sharply.

A system of mathematically verify faked photos will help detect amateurish attempts at fakes, but have the converse affect of making truly sophisticated fakes even worse by creating a way to tag the legitimacy to the faked photo.

Re:Seamless Math Next?

[SEWilco]

You thought he did not integrate it seamlessly?

Re:Seamless Math Next?

[Anonymous Coward]

There is nothing so amorphous and heterogeenous as a photograph. No two pictures taken of the same subject will be exactly alike. Assuming that a 'fake' will be for a reason; and further assuming that this reason will be an addition or deletion or
mixture of some of the same, this will further excacerbate the differences. A 'foto analyst' will probably have no 'reference foto' of any particular subject matter. Such 'reference will have to come from the same camera as the study foto on order to reproduce reasonably the same shot for many reasons
including but not limited to lens defects, exposure program duplication, timeing, operator shake of the camera, and other reasons...
For these reasons, the idea that some mathmatician can calculate a fake with statistics is patently fatuous! Any believer in this would be a fool. Many fools exist in governments however. They are called 'civil service'. Many more are in private industry. These worthies are called 'executives'.
I know of what I speak. I have a minor in math
and over 190 credit hours on my Bachelors in Engineering. This record includes statistics. I love to work in digital images. Instead of looking for 'checksumsin the dark' which is for dorks, look instead for small cut lines to detect cut in images. Most image parts should kind of fade into the other parts of the image. If this border is sharp, then that part may be added, especially if the focus is much different for similar image parts in other parts of the larger image. For this one requires a human so far, and still the results will be inexact, and would not hold up in a fair court where 'beyond a doubt' is the criterion of proof. For some civil cases where 'preponderance of evidence' is the standard, this still would be unreliable unless corroborated with other evidence.
In the old days of typewriter forgeries, one could literally 'fingerprint' the typewriter. Now
one can take a pic, take another. Cut out an image on a 'blue screen' and digitally add it to the other image in a third working final. Then one could print the third on a high definition printer. This print can be run through a color scanner at a slightly lower resolution as the original, and the process done over and over until
the result looks like a slightly degraded original shot. Here the repeated scannings will add just enough fuzz around all the image parts as tho make the whole 'all new again'!

Re:Seamless Math Next?

[fcrick]

I totally agree - this is like saying spam filters will completely block spam - and they just don't. Even with all the extra tools that aren't available to forgery stoppers (people reporting spam, spam traps, IP blocking) we still can't filter out the SPAM. What this guy seems to be claiming is that a photo that fools the software can't be made, even if the software's feedback is used to make the forgery - if he really believes that is the case, that should be the focus of the article.

Re:Seamless Math Next?

[can56]

Nonsense. The Fourier transform of a series (1d) or image (2d) depends on the original signal -- there is no such thing as a charactertistic power spectrum in natural images. Take a picture of 1) a black and white tiled floor, and 2) a dog. The power spectrum of the FFTs of these two "natural" images will look somewhat different ;-) The Popescu/Farid paper simply shows there is a difference (in the FFTS) between original and resampled images.

Re:Seamless Math Next?

[freshmkr]

there is no such thing as a charactertistic power spectrum in natural images.

You can fault me for not reading the research under discussion, but before you make this claim, I recommend a glance at these papers:

D. L. Ruderman and W. Bialek, ''Statistics of natural images: scaling in the woods,'' Phys. Rev. Lett. 73, 814-817 (1994)

J. Huang and D. Mumford, ''Statistics of natural images and models,'' in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (IEEE Computer Society Press, Los Alamitos, Calif., 1999), pp. 541-547.

It's true that individual images will have divergent power spectra (otherwise what, exactly, would Torralba be doing in his work, e.g.?), but on the whole, natural images reflect certain statistical tendencies (again, what, exactly, would Torralba be doing otherwise?). Whether this is discernible enough on the level of single images to help detect forgeries is something I can't reliably say, this not being my research focus. But it wasn't as bad a guess as you thought.

--Tom

Re:Seamless Math Next?

[Paradise Pete]

LOTS of people are still leaving fingerprints as clues at a crime.

It's not the average shmoe that's the problem. It's government inventing the news. The better they can do it, the more they can control.

Re:Seamless Math Next?

[Anonymous Coward]

This shouldn't be too hard for a genetic algorithm. The GA would greatly reduce the time needed to find the right pixels.

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...

Re:Actually...

[Anonymous Coward]

>I can tell you that the hardest thing is knowing when something's right.

I think you have it the other way around. Knowing when something's not right is _easy_. One can always tell when a drawing or picture doesn't look right. It's amazingly easy to tell a drawing or montage was made by a student compared to one of a master.

What's hard is knowing _why_ it's not right.

v

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

[Anonymous Coward]

GIMP -> Filters -> Statiscs -> Match layer to bottom

Re:Self Defeating

[slavemowgli]

True. And for the same reason, encryption is useless - a determined enough attacker could still break it (assuming he invests enough time / energy / computing power), right? ^_~

Re:Self Defeating

[pcmanjon]

"True. And for the same reason, encryption is useless - a determined enough attacker could still break it (assuming he invests enough time / energy / computing power), right? ^_~"

Unless it's shadow encryption.

In that encryption it's 1 way.

It's how linux passwords work... of course you can't "decrypt" in that form of encryption and thats why breaking a UNIX pass by anything other than brute force is impossible.

This is confusing but it basically works by making a hash..

Taking a password: dividing it by 3... dividing it by 2 dividing it by 7...

then when you type a password it uses the same dividing process and checks the "hash" for if it matches.

This however, also makes it so more than 1 password could work... since there can be multiple results... but it's almost unbreakable

Re:Self Defeating

[John Biggabooty]

I don't know what the big deal is. Old style film photos can be fiddled with, but digital photographs cannot be faked, and digital photography is taking over.

Re:Self Defeating

[T-Ranger]

Two things:

1. Password "hashes" are just that. Special types of hashes known as cryptographic hashes, but hashes none the less. They are not encryption in the traditionl sense of the word.
2. It is, in a sense, possible to "break" a crypyographic hash. You simply put all possible inputs into the hash function and record their outputs. Indeed, a service that can return the input for a given crypt()ed output was recently featured on slashdot. It only worked over a limited range of possible inputs, but over the very likely range of inputs (alphanumeric, <10 characters, IIRC). Dictionaly attacks on password files are common, both by sysadmins and crackers alike. It is only a matter of determination (and storage capacity) to produce such a table for all possible inputs to crypt(), and its modern, stronger, friends. Now, you dont necessaraly have the "correct" input returned when you do such a search, but you get a correct enough input.. And for cryptographic hashes a correct enough input is correct.

Interesting question: how long would it take for Distributed Net to search the entire hashspace of MD5, SHA-1?

Re:Self Defeating

[swillden]

Unless it's shadow encryption. In that encryption it's 1 way.

Encryption is reversible, by definition. You're talking about hashing. Secure hashes are useful, but they're not encryption (at least not by themselves, you can combine a secure hash with some other techniques to make a reversible encryption algorithm that's likely to be reasonably good).

thats why breaking a UNIX pass by anything other than brute force is impossible.

Only as long as the hash truly is one-way. The only way to be completely certain is to discard sufficient entropy, but you can only do that if you have enough entropy (long enough password) to begin with.

Re:Self Defeating

[aka-ed]

Please, someone mod this either "troll" or "funny" (I advocate the latter), lest others take it seriously.

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).

Re:Self Defeating

[flonker]

Actually, it can fairly easily be proven that popular cryptographic hash functions are NOT collision free.

Let's say the hash algo produces a 256 bit hash. That's 2^256 possible combinations. Let's say we hash 2^256+1 items using our algo. We are guaranteed at least one hash collision.

Of course, doing this would take an extremely long time (ie. long past our lifetimes), or a revolutionary advance in computing (ie. quantum computing).

[x] No Karma Bonus, because this is off topic

Re:Self Defeating

[Anonymous Coward]

This is one of the efforts to find displayable proof of the existence of Saddam's WMD.

Re:Self Defeating

[hunterx11]

They are already trying to find MD5 collisions.

Re:Self Defeating

[John Sullivan]

"Search[ing] the entire hashspace" doesn't seem to make a lot of sense - a hash converts an arbitrary sized document into a (often smaller) fixed size value. So there are in theory an infinite number of documents that correspond to each hash. (So none of the following probably means very much until you clarify what you'd be searching for, in what, and how.)

There's an interesting section at the beginning of Applied Cryptography which uses basic physics principals to show that simply counting a 256-bit counter through all values would take far more energy than is available in this universe. Even a 128-bit counter is in trouble purely on energy grounds, and would in any case run out of time in most universe models. Producing all possible key values is fundamental to a brute force attack, and the above analysis said nothing about actually *using* the counted values for anything.

Of course, counting possible hash values gets you nowhere unless you can come up with some cryptanalytic attack which given a hash value, tells you something useful about the documents which could have produced it. For the currently recommended hashes we're nowhere near there yet, and indeed even old hashes were designed to defeat this as best they could, and new hashes do it better.

Now, if you want to go in reverse and see if you can find at least one document that hashes to each possible value your situation doesn't improve since current hashes are designed to defeat deliberate attempts to "work through the process" to attempt to influence the output by tweaking the input - that's how collisions occur. Even if you try to brute force the document space, you have to store the results so far to know what your progress is, which is a truly monstrous (and physically impossible in this universe) amount of storage. And with no way of carefully choosing the input your hitrate on the unexplored parts of the hashspace suffers exponential decay, causing the search to get slower and slower... (which actually doesn't change the status of the problem from impossible at all.)

But once you've finished I'm sure the manager of The Restaurant at the End of the Universe would happily buy your dinner for you.

So?

[lordmoose]

So they've developed a way to find the pattern inherent in natural pictures. And they detect when that pattern has been disrupted.

It will only be a matter of time before someone who manipulates a picture feeds the algorithm back into the picture to make it "natural" again, complete with changes.

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?

Re:Just a thought...

[Alsee]

"model that captures the mathematical regularities inherent in natural images."
I wonder if they've considered the potential applications in image compression?

Actually the quote is incorrect. They have captured a regularity introduced into certain kinds of manipulated images - a regulatity that does *not* exist in natural images. A natural image appears smooth and random under their technique, not finding any information to compress.

Interestingly, I guess you possibly could use their technique to compress images that have been rescaled and/or rotated. However when you compress images you usually want to use something like JPG, and JPGing wipes out exactly the information this technique compresses. The methods fundamentally conflict.

So it would rarely be useful. Acutually I just realized something - even in the rare cases where it would be useful, if you're going to modify a format to handle something like this you'd be vastly better off explicity storing this rescaling/rotation information in the compressed file at the time the image is first rescaled/rotated. That would be vastly more efficent than attempting to detect and invert a scaling/rotation function after the fact. Basicly you'd have an image format with built in commands for rescaling/rotation and the file would be somewhat smaller when thoes features were used. An interesting idea, but it doesn't really seem worthwhile.

-

Re:Just a thought...

[IceAgeComing]

You're talking about the time-space tradeoff that crops up repeatedly in CS. You can make files smaller (not in all cases, but especially if you're stretching an image) if you're willing to include image transformation commands as part of the file. You pay for this by requiring that the transformations be recomputed every time the file is rendered.

It's a bit like the choice between compiled vs. interpreted code. An ASCII file of instructions to be interpreted can be much smaller than the equivalent program compiled into machine code.

I had developed a foolproof algorithm...

[Anonymous Coward]

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

Re:I had developed a foolproof algorithm...

[Anonymous Coward]

Funny, I had one similiar. Let me look....... Here it is:

-rwxr-xr-x 1 user group 289729 Apr 27 2003 where_s-waldo

Mod Parent Up

[Anonymous Coward]

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

This cracked me up!

Damn

[lustucru]

I couldn't cheat on msn-girls with my "almsot genuine" pictures anymore...

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.

Random set of pixels?

[MedHead]

I fail to see how an image can have a "random set of pixels" as the article suggests. With color and brightness balancing, the pasted areas of an image blend in with the rest of the image (especially when the artist uses Photoshop's clone brush to combine the two images). This sounds to me like the computer could very well throw out false positives for images that have extreme color or brightness differences.

Re:Random set of pixels?

[Anonymous Coward]

This sounds to me like the computer could very well throw out false positives for images that have extreme color or brightness differences.

It's a computer. I'm sure nobody's expecting its results to be useful or accurate much of the time. The point is, new toy.

Re:Random set of pixels?

[Anonymous Coward]

W/o actually reading the dude's papers, i would guess that the algorithm compares sections of the image to itself, statistically matching those sections. So if one section was supersaturated and the rest was undersaturated, then warning bells light up. Cloning tool does blending great, but generally 2 images of different subject differ in subtle variations of saturation and contrast. I don't see how a filter to match one layer's levels to another won't fake out their algorithm. And i wonder how the algorithm handles mixed lighting scenarios (soft light/ hard light as well as different coloured light).

Once someone posts the link to FAKEIMG.EXE i guess we'll know more.

Re:Random set of pixels?

[MedHead]

From the NYT article:

Stretch [an image], and image-editing software like Adobe Photoshop will assign the picture's original pixels to every other slot in the new picture. That leaves 100 pixels "blank," or without values. Image-editing software fills in the gaps by examining what their neighbors look like.

What happens if an image pasted on to another is not transformed? What if it's not stretched, but shrunk?

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.

Doesn't this make this algorithim useless, since digital cameras primarily use JPEGs? And 90% quality trips this algorithim up?!

Re:Random set of pixels?

[Johnathon_Dough]

Different digital cameras capture RGB information differently, so, even though on your screen you could be looking at something that appears visually similar, the balance of the Red Green and Blue Channels would actually be different, especially noticeable in the dark areas.

Another is film grain, not as prevalent in digital cameras, however, with film cameras, grain is varied based on film speed, manufacturer, light level, and other situation. When you comp together images from a variety of sources, one of the first things you do is equalize the grain, so that you do not get this dispariging result across the image.

I already said this in a previous response, but, from the article, this appears something to catch the talented amateurs, and not someone who really knows what they are doing.

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?

Re:Okay, but here's my question.

[BillX]

Conversely though, I wonder if compression artifacts could be one of the very things the algorithm looks at. Your digital camera takes a picture, compresses it with JPEG and produces subtle (to the eye) artifacts. Now start moving things around in the image (lets assume for the moment you're not even grafting in UFOs), and now the artifacts have moved with them, and can be identified as not being where the JPEG algorithm would put them. Of course, to make it look convincing, you've probably used some smoothing, airbrushing, feathering etc. filters, which will have obliterated artifacts entirely in some places. Finally, suppose you do graft in a UFO--where does this picture come from, and does it have compression artifacts of its own? Even if it came from the same camera, differences in the images mean differences in the compression. But the UFO's probably been resized, rotated, etc., further messing up the artifacts. Of course, just like when you move stuff around in the same image, they're not going to match up anyway. I think this would be a very easy thing for an algorithm to look at.

Of course, you could just save the doctored foto itself as a low-quality JPEG - but a) if it's going in a newspaper, do you want low quality?, and b) you're still just adding your own artifacts to the original - it's going to be some pretty severe compression before they're undetectable.

Re:Okay, but here's my question.

[drinkypoo]

Compress your JPEG with three different encoders at subsequently lower quality settings until you reach the quality setting (and file size) you're looking for, and the artifacting will be effectively untracable. Unless there has only been one predictable compression pass since the manipulation, the lossy nature of JPEG should be sufficient to make it impossible to track the chain of encoders back to the source image data. (It's always impossible to achieve the original data, that's what lossy means, but you can likely get some good information about what it was encoded with by examining the resulting file.)

Re:Okay, but here's my question.

[asavage]

It would be easier to just use all raw images with no compression artifacts. After the fake photo is complete then it can be compressed if desired.

Re:Okay, but here's my question.

[GlassHeart]

Compress your JPEG with three different encoders at subsequently lower quality settings until you reach the quality setting (and file size) you're looking for, and the artifacting will be effectively untracable.

In a modern court, the defense doesn't have to prove that you've doctored the picture. If they can prove that you've run the image through three different encoders (rather than coming straight from the digital camera), they will likely establish reasonable doubt that the image is genuine.

Re:Okay, but here's my question.

[drinkypoo]

There are entirely valid reasons to use multiple compression engines. For instance you might do your editing in one program, and then optimize the image in another.

Re:Okay, but here's my question.

[Anonymous Coward]

Hany presented these techniques at the Information Hiding conference in May in Toronto; I was present and am sort of an expert in this field.

Yes, JPEG compression AFTER tampering with the photo will probably obliterate the tampering. There's no easy way around that.

As you say, almost all digital cameras do JPEG compress their pictures. If you subsequently splice two images together and then JPEG compress again to try to obliterate the evidence of tampering, you might be exposed because the two amounts of double JPEG compression differ. But if the second round of JPEG compression is harsh enough, or if you use some other technique such as print/scan, you will probably get away with it.

The last paragraph of Hany's paper was very honest in saying that people will easily find ways around these digital image forensic techniques.

Re:Okay, but here's my question.

[GlassHeart]

There are entirely valid reasons to use multiple compression engines. For instance you might do your editing in one program, and then optimize the image in another.

Sure. Proving that multiple encoders went through the image doesn't prove that you doctored the image. The problem is that it's generally your responsibility to prove that you didn't doctor it. If a trusted algorithm cannot certify your final image as genuine (whether due to "valid reasons" or malice), then the image may not be admissible in court at all.

Re:Okay, but here's my question.

[drinkypoo]

I wouldn't expect any digital image to be admissable in court, though. If I were taking photographs for the purpose of producing evidence, I would be shooting them with film. Even a disposable film camera is superior for the purposes of producing photographic evidence to the finest digital camera for this reason.

Re:Okay, but here's my question.

[rgmoore]

In a modern court, the defense doesn't have to prove that you've doctored the picture.

Which is only relevant if the faked photograph is being used by the prosecution in a criminal case. There are plenty of other uses for faked photographs. Somebody might want to defame somebody they don't like (like the faked picture of John Kerry standing next to Jane Fonda at an anti-war rally) or manipulate the public into supporting some cause that they otherwise wouldn't accept. In a situation like that, it would be up to the people who disagree with the manipulator to prove that the photo is a fake, and it's quite likely that the only court that would matter is the court of public opinion, which doesn't have such neat and tidy rules.

Re:Okay, but here's my question.

[vadim_t]

The simple solution for that would be to use all RAWs for your modification, and then save as JPG. Then there will not be artifacts out of place, and additionally JPEG compression will probably hide small mistakes you make with photoshop.

I suppose that if somebody seriously wants to falsify evidence, they'd at least spend some effort on doing it well.

Re:Okay, but here's my question.

[GlassHeart]

Somebody might want to defame somebody they don't like (like the faked picture of John Kerry standing next to Jane Fonda at an anti-war rally) or manipulate the public into supporting some cause that they otherwise wouldn't accept.

In such cases, the damage is almost always done at the point of initial publication, and no amount of proof matters. A significant percentage of US citizens still believe that Saddam Hussein had something to do with 9/11, and that didn't even involve a sensational doctored photograph. All the technology available is not likely to help this sort of thing, unfortunately.

Re:Okay, but here's my question.

[GlassHeart]

Even a disposable film camera is superior for the purposes of producing photographic evidence

...unless there was somehow an algorithm that can conclusively (as conclusively as DNA evidence, for example) prove that an image has or has not been doctored. Even if it ultimately cannot be used in court, it can provide investigators with confidence that they are believing in the right evidence.

Re:Okay, but here's my question.

[tricorn]

Unless the camera adds a secure signature to the image (which, of course, means that the original image as the camera produced it has to be available, no cropping or contrast adjustments or color balancing or re-compressing it.

Of course, then you could make a "validated" fake just by creating a fake and taking a picture of it with a secure camera. Ultimately, the only way an image should be admissible is for the person who took it (or who set up the equipment and removed the images, say in the case of a surveillance camera) to testify that they took the picture, it is an accurate representation of reality, and/or testify to the chain of custody and that no one could have tampered with it.

Re:So...

[Anonymous Coward]

Funny this rude photo is actually on topic! I've always wanted to know if that was real or not!!!

This is going to throw M2 a huge curveball...

huh?

[dugger]

Am I wrong or would such software only detect such changes in the original altered image file and not the unlimited number of copies that could be made?

If you, say, paste a turgid phallus into Bob Dole's hand to replace his famous pencil, you'll (stretch, shade, twist, or otherwise change) a lot of pixels, but if you take a screen shot of said altered image then paste it into your image editor as a new image, wouldn't it have an entirely fresh and consistent pixel 'construction'?

Re:huh?

[slavemowgli]

You are wrong - in fact, your idea of how digital image processing works seems to be entirely warped. (This is not meant as an offense in any way, BTW.)

Re:huh?

[doublebackslash]

No.
The Pixels are not stretched, they remain squares, the color is inteigently manipulated by Photo Shop or whatever, but the pixels are not moved.
The picture on a digital medium would be a perfect replica (or almost perfect for lossy compression) of what you were working on.
The 'pixel constructin' is refering to the way that the colors are aranged on the image, particularly on the edges of objects. These colors, and their locations will transfer to paper, jpeg, gif, etc. So long as the replica is of suitable quality, there will be traces of the forgery.
A 'consistant pixel construction' just means that all the colors make sense if the picure was real and taken all at once.

Re:huh?

[Lehk228]

not at all, if you took a screen shot the resulting image would be the same, an image is simply a two dimensional array (or compressed equivilant) of RGB values, copying that array results in an exact copy.

Utterly Devoid of Content

[brauwerman]

Can someone post a link to a real article on the topic? An article with... content?
Editors, please stop posting Press Releases.

Re:Utterly Devoid of Content

[nusratt]

did you read *both* linked articles?

Re:Utterly Devoid of Content

[Anonymous Coward]

I don't know if you actually read the article, but if you'll look closer, you'll notice the press release contains links.

Old Problem

[Detritus]

I've seen picture postcards from the early part of the twentieth century that were obvious forgeries. Things like a picture of grandpa with his six-foot long trout. The Soviets used to erase non-persons (arrested or dead) from published photographs.

Re:Old Problem

[Anonymous Coward]

http://elmo.academyart.edu/study/ph101/Landweber_I mages/31.jpg

Description / demo

[slavemowgli]

I wish there was a more detailed description of the algorithm used and/or an online demo available...

Re:Description / demo

[CHICK543]

read the NYT article. It has a few good details.

Re:Description / demo

[nusratt]

"I wish there was a more detailed description"

http://science.slashdot.org/comments.pl?sid=115590 &cid=9789556

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.

Re:The unfortunate thing

[Anonymous Coward]

it appears he will be publically presenting a paper covering the fakeness-detection algorithm

Thank $DEITY for that, because the article itself was frustratingly light on detail.

Actually... it had *none*. Was I the only one expecting a 'page 2' when it became apparent that nothing technical was going to appear on page 1?

Re:The unfortunate thing

[Anonymous Coward]

Re:Parent. Okay, I'm a ****; was that NYT link there before? At any rate, at least the second article goes into some depth.

Re:The unfortunate thing

[fermion]

If you really want to get paranoid, the more likely situation is this. The US releases a product that is purported to be effective at identifying manipulated photographs. This method works on all photographs manipulated using commonly available technologies. The product is widely available so that independent parties can run thier own test.

The US releases a photo, say of nukes in Iraq or Michael Moore molesting a child, and allows any interested party to examine it. The photo passes all test at authenticity, including that of the said product. The photo is reported as passing the test, and therefore could be authentic. Fox news reports the photo as 100% authentic beyond any question.

"omg alien pics pls"

[desplesda]

Sod the photo manipulation detection article, does anyone know where I can find a copy of the alien photoshopped image the post mentions? :D

Re:"omg alien pics pls"

[nusratt]

"where I can find a copy of the alien photoshopped image?"

If you're referring to the saucer-over-SF image, why don't you inquire at info@WholeEarthMag.com ?

It will improve the fakes

[bhsx]

People making fakes of UFOs and breasts on the Olsen twins will discover new techniques to overcome the detection. It will always continue in a spammer/spam-filter fashion. I'm sure that we'll soon see tools to automatically take the mathematical principles into account and automatically correct fakes that can be detected.
But hey, I'm pretty sure this one pic of the Olsen twins I got from Kazaa is real anyway.

Re:It will improve the fakes

[polyp2000]

I'd have to agree. It stands to reason that if you can create a mathematical analysis tool that ascertains an images likelihood of fakeness, It is entirely conceivable to use the same mathematical techniques to process an image in such a way that it passes the test.

Re:It will improve the fakes

[-noefordeg-]

Yes. Just what I thought too.

If we increase our understanding of image manipulating it will work both ways. If you know how to do something 'backwards' you'll also be in a better position to do it the other way around.

Re:What kind of digitized photos does this work on

[Anonymous Coward]

I can imagine a class of images where subtle manipulations caused great effects and were not readily distinguishable from compression artifacts.

You can, but statistics can't (at least until filters to statistically match one layer to another are developed).

Re:What kind of digitized photos does this work on

[gl4ss]

..and how will it ever work on to prove fake a simple ufo forgery where you just slap something over white/blue background? sure it would probably detect some mild photoshopping made to spice up a picture to look good(or more 'real life', with unnatural blurs & etc). but the more plain the picture is the harder it would be to detect any tempering by just looking at the photo.

good way to get some pr for some uni though, people like magic like in some bad scifi where they can detect which areas are fake in a picture AND magically retrieve the original parts.

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

Re:Why not *make* it real?

[Anonymous Coward]

Because the resuling image will still have been digially edited. It can detect that.

Re:Why not *make* it real?

[Johnathon_Dough]

Print it on a high quality printer

The short answer is dot structure. All printers, (excepting Dye Sub) use some form of sparying of ink, or layering of screened images(halftone dots) Fancier does not necessarily mean smaller dots, it usually means more calibrateable and more consistent color. What you need is a film recorder, which will transfer your image to a negative or a slide, and currently, there are no digital cameras that will record an image at a high enough resolution for this to be flawless (about 40-100 pixels per millimeter), so you must start with a film camera and scan the image in at a high enough resolution. The typical rule of thumb for this kind of work would be one size up from what your end result would be. So if I wanted to create a faked 35mm slide, I would start by taking pictures with a 4x5" film camera scan and use that.

But yes, your point is sound. Technology will not fill the analog hole. Even in "digital" photography.

Re:What kind of digitized photos does this work on

[Gibberlins]

Good question. Someone else who knows more about it should clue us in. I have taken a couple of image processing classes and know that quite a bit a preprocessing is done before you even get the image out of your digital camera. Noise is one of the biggest issues(getting all the signals from the Millions of pixel elements isn't easy) as far as I know.

O.o never heard of common sense?

[Turn-X Alphonse]

99% of the time you can pick out photoshopped images really easy. If it seems to good to be true it probably is, other wise theres always tiny misalignments here and there

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

Re:A more in-depth article

[Werelock]

Mod Parent UP!!

This link also provides links to his other papers, all in pdf format. I'm on dialup and still downloading, but the intro text for the above page says his algorithm works on TIFF and JPEG/GIFs with minimal compression. Hope someone puts up a demo page sometime soon.

Re:A more in-depth article

[Anonymous Coward]

God. I've just read the first page and it seems like there's an easy way to defeat it. Add noise. Futhermore, as even they suggest, there are certain ways to defeat the algorithm. Downsampling by two (no anti-alias filter) will not produce any periodic patterns. I'm not sure how the method could work on any image which has been lowpass-filtered and downsampled.

It seems to deal mostly with the way most drawing apps resize images. A more interesting (though perhaps doggedly slow depending upon the new size you would want to attain.) method would be to take the discrete fourier transform of the image, resize it appropriately, (* add some noise to the higher frequencies if you are extending the size) and take the inverse transform. I'm willing to bet this has a negative effect on the result, as I beleive (INAEE) that the pixels in the image would have some correlation to all the other pixels.

The real thing that anybody forging images probaby already knows is NEVER RESIZE an image bigger! Even a goldfish brain could figure out that. If you resize the image larger, there will not be the detail expected. I think most human viewers could detect this.

Bleh, its probably too late to post this.

If it has an alien corpse, bigfoot, or Jesus....

[Anonymous Coward]

It's not real. I don't see why we need this

Embed a private key in the camera?

[e9th]

Suppose Nikon e.g., were to bury a private key in their cameras and use it to sign the raw image. Then the corresponding public key could be used to verify the image.

The only real problem I see is preventing someone from reverse engineering the private key.

Re:Embed a private key in the camera?

[Sardonis]

Furthermode, someone can take a picture of a faked picture with such a camera...

Re:Embed a private key in the camera?

[Anonymous Coward]

The only real problem I see is preventing someone from reverse engineering the private key.

You can't reverse engineer a private key, but you might be able to dig it out of the camera itself.

Re:Embed a private key in the camera?

[pedantic bore]

That would work for the original image (at least until someone figured out how to extract the private key). It doesn't help for images that have been cropped/color-balanced/equalized/etc. So to prove that photos you have are "real" you'd need to both keep the original image and a complete journal of every operation you performed on it since. (Then other people could review the journal, confirm that it leads to your final image, and doesn't contain any operations that distort the truth.)

The real drawback of this is that there are already a zillion digital cameras in the field that do not have this feature, and they won't all be replaced with these new cameras any time soon.

Re:Embed a private key in the camera?

[e9th]

The raw image file in high-end digital cameras contains enough information to make "photos of photos" detectable. Try taking such a picture with the same shutter speed, aperture, color balance, and lighting as the original.

Re:Embed a private key in the camera?

[atcurtis]

If a camera manufacturer were to do this in a *secure* manner, it would be quite expensive. The 'signing' silicon would have to be on the same silicon as the CCD element otherwise it would be vulnerable to a simple attack where someone feeds the already doctored image in by (for example) creating a FPGA which looks like the CCD element and interfaces to some data source.

But I was thinking... What would prevent against a 'old-fashioned' fakery? Imagine getting a good high-resolution image of Bin Laden and Bush Jr, making full-sized cardboard cutouts of them, setting them in a set with furniture etc and then taking a picture of that - if the lighting is done correctly it would look good and it would have no artifacts from any image manipulation.

Re:Embed a private key in the camera?

[e9th]

I was thinking especially of cameras used for forensic/investigative purposes. I guess the manufacturer could even generate a seperate key-pair for each camera, stored in a tamper-evident fashion.

Then, if an image was called into question, a factory expert could testify that a particular image was taken by this camera, under these conditions, with these settings, at this time.

Re:Embed a private key in the camera?

[t_allardyce]

I think 3D (Z-axis) cameras are starting to become availiable, you could stick one of those in to make sure it wasnt a flat cardboard cut-out or a picture, also you could stick in an infrared camera to make sure they were giving off the right sort of heat signature (and not that of a hotwater bottle stuck behind them) add to that that all sensors have as near to 360 view as you can get (you cant use a studio or it will appear in the shot) make it a video camera (because thats harder to fake) and get the evidence directly from the cameras tamper-proof self-destructing case and then we're talking :p

Re:Embed a private key in the camera?

[tricorn]

You don't need to keep a complete log of how the original image became the published version. You only need the original image so you have something to check to see (subjectively) if their was distortion in the final product. That can be done simply by looking at the two, without any details of how part of it was lightened, a little red eye removed, cropped and rotated, etc. The original shows "the truth".

You can put the private key in a tamper-evident chip - the chip must still be intact for it to be used as evidence; the public key should also be signed by the manufacturer.

Re:Embed a private key in the camera?

[pedantic bore]

...so you have something to check to see (subjectively) if their was distortion in the final product.

Yes, if you're willing to eyeball everything. Personally, I prefer to let a computer do that sort of thing. (And courts like journals because they remove some layers of subjectivity.)

If I wanted to create photos/vids of a UFO

[WormholeFiend]

I'd make a totally black saucer or triangle shaped balloon, as big as possible on a budget, and fit it with blinking IR and UV spectrum-only LEDs, and then have a few accomplices take photos with both film and digital cameras, along with some other people filming with camcorders on normal, and others with camcorders on Nightshot or equivalent night-vision.

Re:If I wanted to create photos/vids of a UFO

[nusratt]

"I'd make a totally black saucer or triangle shaped balloon . . . then have a few accomplices . . . "

I apologize for being so thick, but I'm not following you. Exactly what would be accomplished by doing this?

Re:If I wanted to create photos/vids of a UFO

[WormholeFiend]

well, the ufo-believer crowd would say "look! it was in STEALTH mode!!! we have PROOF!!!"

it would be too simple to make a balloon that is too obviously trying to attract attention.

and since the ufo believers think they're so good at figuring out hidden stuff, i'd be playing right into their psyche.

MICHAEL, thanks for adding the . . .

[nusratt]

NYTimes story link, which is actually more informative and interesting than Dartmouth's own story. In particular, the instant that I started to read the NYT story, I dope-slapped myself for not having thought of the reverse implication of the technology, namely that it might be used to prove that a contraband image (such as child-porn) is NOT faked (and therefore is genuinely illicit).

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.

Re:MICHAEL, thanks for adding the . . .

[nusratt]

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

I apologize for having phrased it poorly. My point is the following scenario . . .
-- Mr.X is arrested for having child porn.
-- His defense (recently legitimized by the U.S. Supreme Court) is to say, "The images are digital artifice, no actual children were involved."
-- The prosecutor says, "We ran the images through Farid's process, and there's no evidence of alteration. The artifice which you claim is involved, wouldn't have been detectable to the human eye. Therefore, you had no plausible reason to OBSCURE the artifice, since Farid-detectability wouldn't have detracted from the ostensible purpose of having the images. Therefore, your claim of digital artifice is not credible, and I believe the images to be genuine, and so should the jury."
-- Barring a convincing explanation from the defendant, the jury might very well convict, in circumstances where a pre-Farid prosecution would be much more likely to fail.

Got it?

Re:MICHAEL, thanks for adding the . . .

[Secrity]

Damn, now somebody is going to have to write a program so that they can process all of their pictures so that their pictures can be plausibly deniable. Maybe not, as most people that I know edit, crop, convert, or resize picture files when they export them from their camera.

Re:MICHAEL, thanks for adding the . . .

[Alsee]

no plausible reason to OBSCURE the artifice

No. I read the paper (actually I read half and skimmed half). The method only detects certain kinds of modifications. There is no need to "obscure" any modification that it doesn't happen to detect.

In particular it only detects rescaling and rotation. And it can't even detect those if the inserted portion of the image origionally happened to be at more than double the resolution as teh final image. here is absolutely nothing nefarious or "obscuring" about having a large (higher resolution) source image and ending up with a smaller (lower resolution) product.

The use of JPG or almost any other compression format destroys the patterns it detects anyway.

So while the method can detect certain kinds of fakes, it is essentially useless for claiming that an image has *not* been faked. Especially if it is a JPG or other compressed format.

defense recently legitimized by the U.S. Supreme Court

Perhaps I am missreading your intent, but it sounds like you dissagree. The alternative amounts to "criminal possession of fiction".

-

Re:MICHAEL, thanks for adding the . . .

[tricorn]

So if the image is real, you can get away with it by doing a little image manipulation to make the test kick out that it's been manipulated?

Re:What kind of digitized photos does this work on

[LiquidCoooled]

The easy answer to that lies in the original compression artifacts remaining - any new fragment/change will not keep these in a statistically similar fashion, and thats what my understanding of this software is.

Smudging a part of an image would remove these artifacts, and would be near impossible to reproduce - like the paper grain on a canvass oil painting.

light popularistic article

[gl4ss]

******
"With today's technology, it's not easy to look at an image these days and decide if it's real or not," says Farid. "We look, however, at the underlying code of the image for clues of tampering."
- Hany Farid

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.
*****

doesn't really say anything beyond "hey we just look at the matrix code you'll see irregulaties in it and you'll see a deja vu effect and presto you'll know the matrix has been changed by agents! MAGIC!"

probably just checks spectrums or something? how does that provide anything really useful beyond knowing if some parts are just from totally different pictures and even then only if the color saturation and stuff like that are way off?

Re:light popularistic article

[Anonymous Coward]

I never thought I would say this on slashdot, but...
I am working on these techniques! The most successful
ones use Support Vector Machines (a powerful class of
classification algorithm) to distinguish true image from
fake image, computer generated images, images containing
a steganographic content and so forth. This is pretty
interesting. I recommend to check out the articles available
at Pr. H.Farid's group page:

http://www.cs.dartmouth.edu/~farid/publications/

Re:light popularistic article

[gl4ss]

pretty cool stuff, some of the .pdf's contain pictures too that shed some light on the issue(though, in one of them the chair alteration isn't done that well at all it's visible by eye in the original picture as well and as such the magic of showing the altered region kind of disappears). it's standard fare to edit the pictures, zoom them, alter contrast, alter colors and stuff like that before publishing.

but it's still problematic because in the cases where you would need to be able to prove it to be a forgery chances are that it could look like a forgery even if it was real or when you have just a crappy print of the forged image.

photoshop

[sosuke]

of course photoshop implemented currency copyright protection ... but i have some convincing photos of my sell in neal's spacesuit, wonder if they would pass the test? is there a where is the free online photo scan?! and does this work on tabloid pictures cause damn they are freaking horrible!

practical use

[Slothy]

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

Re:practical use

[iNetRunner]

What is that? Positively authentic or positively a forgery?-)

*Oh well.. Given people, those kind of photos are quite likely to be true..*

Re:So much for my faked photo of Rumsfeld & Sa

[IamScared]

Worse.

Now my picutes with girls wont cause so much envy in my geek friends...

Bye Angelia Jolie...
Bye Liv Tyler...
Bye, bye..
I like you them all.

Shame on you, heart-breaking algorithms!

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.

for those who want more information

[flynt]

Anyone who wants to know more can read the actual papers here: http://www.cs.dartmouth.edu/~farid/publications/ ...papers 1,2,4, and 5 seem to be what this research is about, but others probably cover it too.

Heh

[Anonymous Coward]

Maybe so, but that mod looks like it will be a hell of a lot of fun!!! :P

NY Times Article Text

[Billobob]

for those who care about it...

For Doctored Photos, a New Flavor of Digital Truth Serum By NOAH SHACHTMAN

Published: July 22, 2004

From the material found on his hard drive, Bryan Sparks of Springfield Township, Ohio, seemed guilty when he was arrested in 2002. The sexually explicit pictures of minors appeared to put him on the wrong side of child pornography laws. But at his trial this spring, Mr. Sparks was acquitted because no one could tell for sure whether the images were authentic or just clever digital forgeries.

Mr. Sparks was eventually convicted on a separate charge of rape and sentenced to life in prison. But the uncertainties that surrounded his case, and others like it, are driving researchers to develop software that can automatically figure out which digital pictures are real and which ones are fake.

"It used to be that you had a photograph, and that was the end of it - that was truth," said Hany Farid, an associate professor of computer science at Dartmouth College who is a leader in the field. "We're trying to bring some of that back. To put some measure of guarantee back in photography."

At stake is more than the fate of possible child pornographers. The United States military has become increasingly reliant on digital images from drones and satellites to give soldiers a sense of the battlefield. Law enforcement officers routinely use digital cameras to photograph crime scenes. Newspapers and magazines are now dependent on digital photographs that can be easily doctored.

Over the last three years, Professor Farid and his students have become experts at forgery, making hundreds of images that look authentic but have in fact been digitally tweaked. License plate numbers are changed. A single stool standing on a checkerboard floor is suddenly a pair of stools. Dents on a car are wiped away with a few mouse clicks.

The skillful tampering disturbed the images in ways that the human eye could not detect. But Professor Farid says his algorithms can spot them and sound the alarm.

For example, when two images are spliced together - like the picture of a shark attacking a helicopter that has circulated around the Internet in the past few years - one or both of the original pictures usually has to be shrunk, enlarged or rotated to make the pieces fit together. And those changes, no matter how artful, leave clues behind.

Take a picture that is 10 pixels by 10 pixels, for a total of 100. Stretch it to 10 by 20 pixels, and image-editing software like Adobe Photoshop will assign the picture's original pixels to every other slot in the new picture. That leaves 100 pixels "blank," or without values. Image-editing software fills in the gaps by examining what their neighbors look like, and then applying an average. To oversimplify, if pixel A is blue, and pixel C is red, the blank pixel B will become purple.

This kind of averaging becomes "pretty obvious" after some analysis of the image, Professor Farid said.

In tests on several hundred doctored photos, this technique for detecting changes proved to be virtually foolproof if the picture quality was high enough. Uncompressed TIFF image files, which contain enormous amounts of data, were like an open book to Professor Farid's team.

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.

Jessica Fridrich, a research professor in electrical and computer engineering at the State University of New York at Binghamton, is approaching the fraud problem from the other side. She is trying to figure out who took the digital picture in the first place.

Just like the rifling in a gun barrel leaves a distinctive pattern on the bullets it fires, a digital camera has a signature of sorts. Today's digit

Re:NY Times Article Text

[Thomas Shaddack]

Now, she is trying to embed a bit of the photographer in the picture, too. The patterns in the iris - the colored part of the eye - are at least as distinctive as a person's fingerprint. With money from the Air Force, Professor Fridrich is designing a camera that takes two pictures at once: one though the camera lens, and a smaller one of the photographer's iris. The iris image along with the time and place of the photo session would then be compressed, encrypted and instantly hidden within the larger picture just taken.

If that encoded information is someday tampered with, or missing, then it is obvious that the photo can no longer be trusted.

Countermeasure against this is easy; just print the image with sufficiently high resolution, then photograph it. Voila - a fresh iris image is encoded in the file.

Why not just digitally sign the file in the camera itself? Alternatively, if for forensic purposes, connect a cellphone to the camera and wirelessly get a timestamp from an authorized timeserver, and register the image's signed hash there. Any subsequent tampering is then very easy to detect.

Re:So much for my faked photo of Rumsfeld & Sa

[Anonymous Coward]

Hang on to those 15 minutes of fame by having Elvis shaking hands with the alien then.

Over at the Whitehouse...

[MosesJones]

GWB: Lets just creatify some photographs of those Iraqi WMD

DR: Great plan Mr President

GWB : Dang! It won't work, they'll provify its a fake

DR: Not till after the election George... not till after the election.

More About Investigating Digital Images

[rpiquepa]

I wrote about this technology a while ago, in "True or False? Investigating Digital Images." A keypoint is that the Dartmouth College team thinks that their technology, or a similar one, will soon be incorporated in the U.S. legal system to authenticate images. At the above link to my blog, you'll also find an analysis of a forged image and more references, including the full research paper published by the IEEE Transactions on Signal Processing journal.

Re:More About Investigating Digital Images

[TubeSteak]

The NYT article talked about a porn case where the prosecution couldn't prove the images were orignal & not fake... If the method of determining validity involves analyzing patterns etc. what's to keep someone from writing a program that will randomly change 0's & 1's? If this program doesn't significantly deteriorate/damage the original image, it'll look fine, yet it will also be detected as 'forged'. Once that happens, you won't be able to authenticate the image, nor conclusively declare it a forgery.

I realize it isn't as simple as flipping 0's & 1's but you get the general picture.

Photoslop

[Anonymous Coward]

You mean that all the nude Brittany, and Anna K is fake?

good background and intro, little details

[randyest]

Farid and his students have built a statistical model that captures the mathematical regularities inherent in natural images. Because these statistics fundamentally change when images are altered, the model can be used to detect digital tampering.

That's pretty much all the detail on the method to detect image altering. Seems reasonable, but:

1) How many real photos deviate how far from the statistical "norm" (i.e., how likely are false positives when checking for alteration?)

2) How long before there are tools that can inject the proper (expected) statistical characteristics into a faked image?

These are not addressed in the article. Anyone have more info?

Re:good background and intro, little details

[Anonymous Coward]

I actually know Farid, and I've seen him present earlier versions of his work...

If I remember correctly, the 'statistical model' part is looking at certain distributions of higher-order statistics of the photos... and your point (2) has been brought up a couple of times, and I guess you could modify a photo to have the 'correct' distribution of those statistics, but you have to modify it so that, at the same time, it also looks like the original photo... and that, as far as I know, would be _hard_.

Re:good background and intro, little details

[NoYes19]

How is it on false positives? I don't see how any such algorithim that can catch quality photoshops would not produce false positives on images thatm for example, have strange lighting.

Re:good background and intro, little details

[Johnathon_Dough]

2) How long before there are tools that can inject the proper (expected) statistical characteristics into a faked image?

There is one, it is called photoshop.

From all of this I have read, here are the steps to fool this process.

1-Manipulate image, smooth any areas with disparaging noise 2-New layer, set layer to overlay fill with 50% grey.
3-Filter ->add noise (pixel ratio based on image resoltuion)
3-Filter->Gaussian Blur (usually .2-.7 pixels)
4-Curves endpoints in 10% top and bottom
5-Filter->noise half what you did before
6-Filter G. Blur again

This should give you a realistic looking film grain (settings vary based on image resolution) with practice you can match a film grain from a supplied image. By creating it on a layer, you can then brush it in on your comped image to family the various bits together, and as it adds some irregularities that appear random, it would probably fool the software

What no one seems to have mentioned though, is most fakes are spotted immediately do to light and shadow error.

Re:good background and intro, little details

[Anonymous Coward]

RTFP. 0.5% false-positives on their training data, ~1.5% on their test data... More to the point, the process is (of course) susceptible to certain kinds of manipulations... but it does really well with natural phenomena, even ones like 'strange lighting.' It really is picking up on statistical regularities that result from actually taking a photo (no matter what it is of) vs. those that come from, say, manipulating pixels in photoshop in a mathematical (that is, regular) way, no matter how subtle to the human eye.

reverse-engineering?

[chachob]

perhaps the technique could be reverse-engineered to allow the forgers to know whether or not their images can be detected as forgeries, and use this information to enhance their forging techniques to evade the detection tools...

Using the fake picture scanner backwards gives....

[nmg196]

Using the fake picture scanner backwards gives you a proofing tool to make sure you've perfected your fake picture before it's released. It won't take too long before these detection tools get on to eMule - and once there there, everyone will be able to fine tune their picture until the proofing tool no longer thinks it's a fake... Once in this state - they can release it knowing that the experts will be on *their* side. So such tools will have little use... It kind of works in the same way that you can use unix 'crack' to make sure your users have secure passwords.

Time for a Genetic Algorithm?

[sameb]

If there is an algorithm that can detect whether or not a photograph is forged, wouldn't it be possible to have another 'competing' algorithm that randomly altered bits and used the detection-algorithm as its success scale? (The better it becomes at not being detected, the better the genetic algorithm is doing.) Seems that that'd work fairly well, and would apply to lots of other technologies that require seamless photograph overlays.

It's still not fool-proof

[Digitus1337]

If someone edits an image and then takes a screen shot of that image being displayed then there is no underlying code to tell anything about it one way or the other. The same kind of a thing can be done by printing out an editted photo and scanning it in again. It's just a matter of not being lazy.

please please!

[mqx]

What I'm waiting for now is someone with time on their hands to build a crawler using this tool that goes around the internet and looks for "fake" images, producing some information about:

* generally: how many images are fake, and what are the statistics breakdown on .com, .net ... etc by zone.

* what about the statistics on news sites? how about we have an action group that "keeps the bastards honest" by regularly running the tool across news, government and other important sites.

* how about porn images, what's the fake % there?

* how about a firefox plugin that automatically tests images as I'm browsing, and puts a watermark over those that it finds are fake.

Any more ideas?

Lots of great applications for this. Please write the tools ASAP.

Re:please please!

[sosuke]

i second the firefox plugin, thats enough for me, and it could send back statistics on the images, and as for the porn all you need to do is open it in firefox! :D

Re:please please!

[Anonymous Coward]

>> * how about porn images, what's the fake % there?

Not as high as you might think, since you'll have a real photograph of fake things...

OK

[6800]

Ok, I'll get the picture :-)

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.

Inrcreasingly more difficult to convince whom?

[heritage727]

increasingly more difficult to create convincing digital forgeries

It only has to convince a jury. What are 12 random citizens more likely to believe, a complicated mathematical formula or a photograph they can see with their own eyes?

Re:Inrcreasingly more difficult to convince whom?

[MedHead]

It only has to convince a jury. What are 12 random citizens more likely to believe, a complicated mathematical formula or a photograph they can see with their own eyes?

Actually, if it was deemed a forgery, I don't think the photo would be presented to jury as evidence. It would be deemed inadmissable.

re: "Embed a private key" -- ooh, GOOD question!

[nusratt]

"Suppose Nikon e.g., were to bury a private key in their cameras and use it to sign the raw image. Then the corresponding public key could be used to verify the image."

Actually, I don't think that'd work, but it raises a very interesting point:
when an UN-faked image is published with any kind of intentional alteration for legitimate purposes -- e.g., digital watermarking to protect the creator's copyright -- could that alteration have the unintended side-effect of making it impossible to prove that the image isn't doctored?

don't believe the parent post since...

[John Harrison]

it's a trap!

Re:don't believe the parent post since...

[zoloto]

wow, /.'s been invaded by farkers! noooo!!

Details

[cmcguffin]

Details can be found in a preprint of the paper here.

("RTFP? I can't be bothered to RTFA!")

Only works if you don't know how to use Photoshop

[Animats]

You don't stretch images; you condense them. The source material must always be higher resolution than the output. Final pass is a very slight Gaussian blur, then an image resize down to the desired scale. This guy is detecting stretched images, which isn't hard.

Or, as artists say, "work big".

Legal issue

[Maljin Jolt]

As I am beginning to understand symptoms of current american legal culture, I can now easily predict some of the future:

- raytracing/3D rendering software will become illegal
- photo editors usage will require a state license
- new color printers will monitor what's printed and sending thumbnails automatically to federal office for approval and archiving
- in all cases old image techology will be outlawed
- painting with oil on canvas will become felony under the "creating unrealistic images" section of the relevant bill

Re:Legal issue

[nusratt]

"As I am beginning to understand symptoms of current american legal culture, I can now easily predict some of the future:
- raytracing/3D rendering software will become illegal . . ."

Well, you're HALF-right, when you say "current"; but whether your prediction comes to pass, depends greatly on what happens in the coming November election. If we can get rid of Inspector Javert -- err, excuse me, Attorney General John Ashcroft -- then there's hope for improvement.

But the problem isn't only in the USA. From what I've been reading, your prediction is equally likely to come true -- I'm sad to say -- in the UK or Australia.

Re:Legal issue

[t_allardyce]

rofl and still anyone will be able to buy an assult rifle..

Re:Legal issue

[lachlan76]

your prediction is equally likely to come true -- I'm sad to say -- in the UK or Australia

One of our candidates (Australia) has said that GWB is the most dangerous leader ever. From the sound of what he says, do you think that we'll be adopting a lot of US law if he gets voted in?

Re:So...

[julesh]

Is this[goat.cx] fake?

Unfortunately (?), the process doesn't work on JPEGs, according to the NYT article.

Re:So much for my faked photo of Rumsfeld & Sa

[Anonymous Coward]

Damn. And there goes my footage of w and mrs. stepford picking their nose and eating it.

What this is not (or what it may not be)

[blonde rser]

There have been a lot of comments so far saying that this algorhythm would be a great help in creating doctored photos. Someone even said that if it passes the algorhythm surly it will pass the human eye.

This may not be the case. There are certainly some features that we are very good at recognizing when there is something wrong: like proportions and white balance changes. I mean we are very good at this so creating an algorhythm to do better than us would be very hard. However there are some features we don't even consider: like in a certain are is every other pixel the perfect average of its surroundings. The NYT article shows an example of the alorhythm using the latter.

I recognize that there are very smart people out there capable of creating very good algorhythms that may work in the first way. But why bother? Someone doctoring an image - up to this point - has only considered perfecting the variants that human eye can catch. I'm not saying what these guys have done is easy or obvious. Quite the opposite. Finding variants that we don't usually requires a huge amount of ingenuity. But correcting these variants won't help fool a person any better.

To all the people who think it's defeatable

[nusratt]

A lot of people have posted about ways to defeat Farid's detection algorithm.

1. I'd never say, "Gee, he's a professor, so let's all just trust him, because he must know what he's doing, right?" But people in his position do tend to be somewhat conservative about pre-publicity, to avoid later looking foolish and damaging their rep. In any case, that's why there's a peer-review process -- which is why I'd put a lot more faith in statements from him, than I would if they came directly from DHS.

2. In the article, Farid himself says, "There is little doubt that counter-measures will be developed to foil our detection schemes. Our hope, however, is that as more authentication tools are developed it will become increasingly more difficult to create convincing digital forgeries."

Re:To all the people who think it's defeatable

[Anonymous Coward]

I think your reasoning is sensible, but I think Farid is overstating his claim.

For instance, removing a person from a picture.
Photo 1: Take the photo with the person in.(ensuring only fixed scenery is behind them)
Photo 2: Take another seconds later when person has gone.
mask & replace the Person in photo 1 with pixels from photo 2.
If done to pixel accuracy how can photo 1 be statistically different from photo 2? (bearing in mind that photo 2 is a genuine picture)

and from the obvious paranoid side...

[zogger]

... they want to be able to pass off their "intelligence" photos as the real mc coy easier, so as to not get busted when they release crap, like the phony fat osama bin laden video(wicked fake) and the (possibly) berg beheading. this technology would help them to establish bonafides with the offical forgeries. We are *this* close to a running man scenario here with them being able to frame people or to alter public opinion with phony video and pictures. If they can create one and have it slip through the checking algorithms, then they are home free.

On the other hand, it would be nice to go back and look at a lot of older photos to see what's what with them now.

Lately there's a smidgen of controversy over some of the mars photos, this would be a great place to use the technique.

Re:and from the obvious paranoid side...

[aminorex]

the fat osama with the funny nose was a hoot, i admit, but i can't see the relevance to digitial compositing, since it was a natural video sequence.

Re:and from the obvious paranoid side...

[zogger]

I was just referring in general to them trying to pass off disinformation as the real deal all the time. I probably could have used better examples. Hmm, the first released zapruder tape with the reversed frames with the headshot in them is a good but crude example.

Re:So much for my faked photo of Rumsfeld & Sa

[Anonymous Coward]

I used to dabble too. Here's a real and the fake photo I did for a group of circus freaks.

the big fuss again...

[l3v1]

Thing is, it is surprisingly easy to detect on an image those areas which were copied from another image. The situation can arise when you could simply just draw exactly the contours of the pasted portions. Things get harder when some areas are changed with parts copied from the image itself (i.e. not from another image), e.g. you cover or replace a part of an image with a part of itself from another place. In this case statistical evaluations can (in some cases) fool you. It gets even harder when the modified image gets through a strong lossy compression, but in most cases this doesn't mean altered areas can't be correctly identified (especially when the whole image gets encoded in the same way, and the statistics of the image areas relative to each other degrade similarly). All in all, what I'd like to say is that this problem isn't as hard to solve as most everyday people from next door would think. One possible solution would be the widespread use of robust watermarking algorithms on the original images. There are some pretty heavily robust methods out there, some of them can even withstand hardprint+copy and more. This thing seems to me like if somebody would come up with the idea that if you eat your hunger goes away.

Wasting Time

[Anonymous Coward]

Why don't they work on something beside trying to prove if Britney Sprears is really naked. How about taking all those brains and putting them into a think take and work on a disease cure?

Re:What kind of digitized photos does this work on

[aka-ed]

This distinction is important, as it seems unlikely for any means to emerge that will overcome these pixel-level distortions.

One wonders whether this will lead to a legal distinction between lossily compressed images and others. While audiophiles have long been ape for lossless compression, not as much a need has been felt for graphics. Where do lossless graphic compression efforts stand? Is this an area where a proprietary standard might lead to big $$$?

doesn't work with jpegs!

[bhny]

from the NYT article- "for now the technique does not work as well with files created in JPEG ...At 90 percent quality, it falls apart very quickly," Professor Farid noted.

doesn't this make the technique almost useless?

Re:doesn't work with jpegs!

[Anonymous Coward]

No, this technique, and other similar authentication methods will make crappy overcompressed jpegs obsolete in applications where authentication is important.

Re:doesn't work with jpegs!

[Thomas Shaddack]

No, this technique, and other similar authentication methods will make crappy overcompressed jpegs obsolete in applications where authentication is important.

If you need authentication, can't you just add a signed hash of the image as a chunk of the JPEG file?

Why not just run the process in reverse?

[Julian Morrison]

Detect those areas that show as fake, and perturb them until they don't. Iterate until the image passes the "authenticity" test.

If you can identify why something looks fake, that information should imply a solution for non-fakeness.

Making Undetectable Faked Photographs Gets Easier

[CedgeS]

To preface, I don't belive this at all. Consider a photograph that has undergone a hue transformation, and is titled George W. Bush has blue face, see photo. It will have all the other properties of a real photo, but will be fake. Another example, consider a photograph of models.

But assuming this is possible, then I can make a fake photograph that is so convincing that they will belive it is true. First construct the photograph, and possibly include markings as to what are the edges of adjusted areas, etc, to aid in developing heuristics. Then search on slight modifications to the photograph until their fake detection algorithm detects the photograph as real.

The above descriped sketch of an algorithm is also a sketch of a proof that their method doesn't work.

In one word...

[Rui del-Negro]

In one word: bullshit.

Just like the "face recognition systems" used to spot terrorists, or the "child protection" software that's supposed to recognise porn.

Not only is the success rate well below 90%, but, more importantly, it spits out thousands of false positives.

And this is without even considering their admission that their technique does not work on JPEG images, even at "90% quality". In other words, they admit it won't work at all in 99% of digital pictures.

These are all tasks that need so much computing power such smart algorithms, that the only system that can perform them with acceptable reliability is a trained human brain.

RMN
~~~

Pop quiz

[arvindn]

Is this photograph real or faked?

Bonus points if you answer the question without using the technique described in the article.

P.S I'm not claiming credit for it; it used to be on his website but isn't anymore.

Re:Pop quiz

[nusratt]

Yes, even to someone who didn't know the parties involved, your linked JPG is obviously fake. But I don't understand your point, unless it was humor.

Re:Pop quiz

[arvindn]

Yes, obviously it was meant to be funny. But I see the humor didn't get across. I'll try to do better next time.

Re:Pop quiz

[t_allardyce]

Maybe its real but hes standing next to a card-board cut-out ;)

3y3 h4x0r j00r ph070

[DarkGamer]

I alter photographs for a living and I think I know of what kind of irregularities he's talking about... when pixellated noisy areas in the channels of an a photo get smoothed out by alteration, like with a feathered rubber stamp tool or by scaling and stretching in Photoshop. If this is what the algorithm looks for it will be very effective at finding BAD photo manipulations. Good technique will yield a visual photographic noise that is indestinguishable from the rest of the image, and will probably fool the algorithm. Just add noise to your finished file and export with some wacky compression.

But then again a human eye could always find bad photo manipulaitons if it knows what to look for.

Color me skeptical.

Embed a GPS!

[silverhalide]

One good way to authenticate an image would be to embed a GPS with a camera. Seeing as this is done on relatively inexpensive cell phones, it shouldn't be too hard. Market the camera as "secure" or "evidence-class" camera, somehow tie the info to the image...

There are security details, but if it is possible to tag an image with WHERE it was taken in a secure fashion, it would eliminate all but the most enterprising forgers.

Re:Embed a GPS!

[t_allardyce]

.. edit the position in the file? the only way to have evidence class devices is if they are locked tamper-proof cases/chips and the evidence is taken directly from them.

Re:So much for my faked photo of Rumsfeld & Sa

[blibbleblobble]

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

Hopefully my photo of Governor Arnie shaking hands with Arial Sharon won't be detected...

They detect resampling

[poszi]

The intro is not accurate. Actually, they are detecting regularities in the forged pictures.

I've looked at their scientific paper and their technique albeit not perfect seems to be very good in detecting any kind of resampling in the image (up- and down scaling, rotating, etc.). When you make the transformation on a grid, the interpolation creates some almost invisible artifacts and regular patters which they are able to find by their analysis. It's difficult to create forgieries without these kinds of manipulation. But probably not impossible.

Re:What kind of digitized photos does this work on

[multiplexo]

Good point. I should have RTFAMA (Read the Fucking Article More Attentively). Of course someone is probably already working on a Photoshop plugin for "correct faked image distortions", which would be handy to have along with some other plugins such as "remove ex-significant other from picture", "reduce glassy-eyed drunken stare", and my all time number one request, "remove drunk friends standing around with dicks hanging out by your mouth when you're unconscious".

Re:So much for my faked photo of Nixon and Elvis

[bennomatic]

LOL, that reminds me of my very first photoshop. It's terrible, but some years ago, a friend of mine and I were in a band (King Lazy Bones), and I was coming up with concepts for an album cover, and this is one of the ideas that I came up with.

If you want to see the original of the Nixon Elvis photo, do a google image search on "Nixon meeting Elvis". And the original I got our faces from (gawd, this brings back memories) is here.

Re:What kind of digitized photos does this work on

[Johnathon_Dough]

I take high resolution TIFF images and "muck with them" every day. Every ad you have ever seen has been digitally manipulated, every video image you have seen. If I am supplied with two good source images, I can create an image that would pass just about any inspection. There are many tricks to get the "irregularities" into (and most times out of) an image. If I wanted to create a fake image, I would do multiple techniques, then "print" the image out to a transparency (slide), or a negative, and have a print made, then scan the print. How would that go through this guy's algorithim?

To me this strikes me as the same sort of "solution" as DRM is, sure it stops Joe Six-Pack from putting Britney Spears' head onto a porn stars body, but it will not stop anyone who knows what they are doing when it comes to digital image manipulation.

Re:What kind of digitized photos does this work on

[GlassHeart]

how can you tell what is a result of manipulation versus what is an artifact of compression and or digitization?

Just a wild idea off the top of my head... while JPEG image compression is a lossy process, it's still a mathematically reproducible process. It might therefore be possible to exhaustively generate the set of originals that could be compressed to the image in question*. If you then apply the detection algorithm to the originals, none of them may pass and you'll still have a valid conclusion.

* Made easier if you know the precise implementation (i.e., which digital camera, which version of Photoshop, etc.) that created the JPEG.

Effect on Steganography?

[Anonymous Coward]

Steganography involves modifying a picture and inserting a message in a way that is supposed to be indistinguishable by even a careful viewer. The point of steganography is to hide the fact that there is a message at all sitting in the picture. But knowing that there is a hidden message is half the battle. Could this technique identify an image as having a message, even if it did not decode the message?

OT but sorta related

[commodoresloat]

does anyone know the real origin of this photo?

Not happening...

[MMaestro]

in the courts. Why? Simple.

Our hope, however, is that as more authentication tools are developed it will become increasingly more difficult to create convincing digital forgeries

Whos to say if a certain method is the right method to use? Just like there are numerous viruses, anti-viruses, and anti-anti-viruses; I think its safe to say there will photo manipulation software, counter-photo manipulation software, and counter-counter-photo manipulation software. Course this leads to programs being compromised and the necessity of making new programs all too often.

Law enforcement applications

[0x0d0a]

In particular, the instant that I started to read the NYT story, I dope-slapped myself for not having thought of the reverse implication of the technology, namely that it might be used to prove that a contraband image (such as child-porn) is NOT faked (and therefore is genuinely illicit).

That's a good point. Hmm. Perhaps a free plugin for Photoshop/GIMP could be released and widely distributed that modifies an image to be in conformance with the model? That'd retain plausible deniability.

My first thought is that JPEG and other lossy compression formats would also throw this off -- sure enough, the NYTimes article backs this up, and says that this is the case (and if someone can come up for a "fix", to determine what probably is a JPEG artifact and what isn't, we have a new and improved post-processor for JPEG decompressors!)

Digital watermarking might also muck things up.

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.

Seeing is believing

[E IS mC(Square)]

"This technology to manipulate and change digital media is developing at an incredible rate," says Farid. "But our ability to contend with its ramifications is still in the Dark Ages."

humm... something similar to the cloning!!? I may be neutral in the cloning issue, but I guess Farid is right here. You dont have to try real hard to fool people. Even obvious fake (Guy on WTC roof before the plan crash!) is believed by 95% of people. And you combine the 'power' of these people to belive in everything they see with the power of faking photographs, you can even make Bob Dole win an election!

Detecting Faked Porn Gets Easier

[Anonymous Coward]

I kinda misread the original title. ;)

I do wonder how effective this will be with porn.

*a sex deprived geek* -JL

It's stupid

[Anonymous Coward]

It's stupid because statistics can be post adjusted
to fit whatever the detection scheme thinks is OK.
There's the same problem in steganography. -AC

So what?

[Anonymous Coward]

"There is little doubt that counter-measures will be developed to foil our detection schemes," says Farid. "Our hope, however, is that as more authentication tools are developed it will become increasingly more difficult to create convincing digital forgeries." [my emphasis]

So what if the casual user can no longer perform forgery? Dedicated forgers will eventually out-model the current forgery detection model, and the cycle continues.

Forgery and forgery detection are a never ending arms race, just two poles along the same axis. Apparently, forgery detection now has the biggest guns.

However, as with any arms race, there is no possible terminus. All this means is that the most powerful weapons become privileged and fall in the hands of the few.

Re:What kind of digitized photos does this work on

[rgmoore]

I strongly suspect that the solution will be some sort of hardware image signing, rather than after-the-fact examination. Canon already offers a Data Verification Kit for their superb EOS 1-DS digital SLR. They don't give too many details, but my guess is that they can attach a cryptographically signed hash of the image data into the file header so that it's possible to confirm the integrity of the data later. Since the EOS 1-Ds can only save data in raw and JPEG formats, and since this doesn't make sense for raw data (which has to be processed to be turned into a viewable image) it seems likely that they have this working with JPEGs.

more difficult, or easier

[oneishy]

If there is a new tool which can identify a photo as fake or not, it simply makes it easier to make convincing fakes! When you can easily identify the sleezy-fake photos you know when you have a good fake. So, if I am in the business of making a fake photo, I know know when it is done right, and undetectible!

The are after all just photos

Re:What kind of digitized photos does this work on

[Anonymous Coward]

I bet you can't pass the "celebrity porn" version of the Turing test.

The whole thing's a fraud...

[MoeDrippins]

...or the guy faked his own photo for the article.

C'mon now. math geek... handsome... math geek... handsome...

Sorry, I'm not buyin' it.

Re:What kind of digitized photos does this work on

[flonker]

The loss is not reversible. Therefore, you would have to guess each bit that was discarded. So, for each bit of information discarded, your number of possible originals would double. Considering how many bits smaller a lossy jpg is than a lossless jpg, this is a lot of possible originals. And they would all look nearly identical.

all this and

[Anonymous Coward]

Farid can't even keep his shirt tucked in for lecture... :s

Easy

[Jozer99]

This is easy, here is some pseudo-code:

Input photograph;
Print "This is fake";

As long as only photographs whos' validity is in question are input, no one will ever be able to prove it does not work.

Re:Easy

[antispam_ben]

Reminds me of the policemen who used a collander and a copy machine (with a "preset message" on the window) on a suspect, pretending these items were a lie detector. When they asked "Did you do the crime" the suspect said "NO" and one cop pushed the [copy] button and the machine spat out a sheet of paper that said "He's lying." The suspect immediately confessed.

Doesn't work on .jpg's

[dpbsmith]

The New York Times article actually says something about the methods used. It basically works by looking at the kinds of operations that need to be performed when interpolating between pixels in manipulations such as changes in size, rotation, etc.

And, "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.'"

Technology

[Anonymous Coward]

I don't remember where I remember this from, I beleive it was a Bruce Coville book, but the quote went something like this:

"As technology improves, so does the technology to fool that technology."

Something to think about I suppose as we move forward.

Am I the only one...

[Anonymous Coward]

who read that as "Detecting Naked Photographs Gets Easier"? I could use something like that.

Attention: OP is a known troll

[Anonymous Coward]

The original poster is a known troll. His "Doom for Columbine" Doom3 mod idea was flamed to death on boards all over the net, but he kept posting that it was really a great idea till he was banned from doomworld.com forums and many others. Check this planetcrap thread, where he goes by the nickname "lexx". But be prepared. You're going to witness a really sick mind in action.

Don't let his low UID fool you. He is nothing more than a lousy internet troll. If u have the mod points, I urge you to mod his posts DOWN! This is a person with a sick mind|disturbed personality. This guy needs help. Urgently.

Re: So much for my faked photo of Rumsfeld & S

[Black Parrot]

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

Yeah, and prove that my photo of Osama and Bert was real!

This camera's authentication feature *is* hackable

[Anonymous Coward]

This can be easily defeated by hacking the 1Ds's firmware. The private key and algorithm used for signing are most likely encoded within the firmware (easy to understand - this camera uses an embedded PowerPC control processor), rather than on a specialized crypto chip. Just pull out the algorithm and key, resign your CRW files and they're authentic.

I have not tried this myself (I've got more interesting things to reverse engineer, plus I don't own a 1D), but sources say this is the way it's done.

The only real solution is to give each camera a private key based on its serial number, store that key in a tamper-proof chip, and provide only the public key with the camera. That would make alterations significantly more difficult, but not impossible (I imagine an organization like the NSA can and would hack the silicon if needed.)

Mod Parent Down

[Anonymous Coward]

> He is nothing more than a lousy internet troll.

Totally devoid of any merit or substance, whatsoever.

The day Planet Crap becomes a website of any value is the day Slashdot is bought by the SCO or the RIAA. All the game developers who ever used to go to Planet Crap have all left because the people there are complete assholes. See for yourself; the last ten "stories" have the phrase "fuck so and so up the ass".

You blatant Idiot.

However...

[Anonymous Coward]

As the ease in making prima facia fake pictures increases it is almost a moot point whether or not a sophisticated analysis scheme can catch them. One use of such pictures is politcal innuendo, which can be generated very quickly by one faked picture and the fact that it's later proven to be false will be overlooked by many. Politicians constantly make wildly unfounded accusations that are easily proven false, knowing that the later "proof" is too late once the falsehood has taken flight. Validation schemes might be useful in legal applications, but I still hold with Marvin Gaye: "believe half of what you see."

This is nothing new

[gwalla]

Batman has already used something like it to determine whether a video of the young Batgirl (before she was Batgirl) assassinating someone was genuine.

is this picture fake?

[Anonymous Coward]

http://urbanlegends.about.com/library/graphics/Aad 4_sm.jpg
if it is could you list what's wrong
I've looked for signs of photoshop and expected to find some but didn't
besides the fact that it shouldn't be true is there something that shows it's fake

Re:is this picture fake?

[antispam_ben]

For some reason, copying and pasting your URL wasn't working, but this link works:
http://urbanlegends.about.com/library/graphics/Aad 4_sm.jpg
Seeing that there are two people apparently cleaning up around an incredibly large human torso skeleton (one person is standing next to the skull, which by proportion would be about six feet in diameter), I would say that either the photo is fake or the human skeleton is fake (as in not a real skeleton, but manufactured for the photo).

Re:is this picture fake?

[whitegold]

A quick search on google shows that this is a photoshop job. Just a damn good one.

It's from Worth1000.com, a website that is worth ANYONE'S time, for it's brilliant imagery. Worth1000 images are all photoshop modified, and by the best. They're rarely intended to "fool" the viewer, rather they are all made to fit a theme, to win a competition.

Am not sure what the specific competition here was, but the following graphic http://www.graphics.cornell.edu/outreach/mastodon/ aerial-views.html clearly shows the same area that is in the giant skeleton pic. Insert diggers and a human skeleton closeup and you're there.

Oh, and to answer your question, I couldn't tell you what's wrong with this pic either, other than the unlikely content. Clearly this was created by a VERY skilled photoshopper. Props to whoever did it.

Matt

Re:What kind of digitized photos does this work on

[mysticgoat]

Where do lossless graphic compression efforts stand?

As completed. PNG uses lossless compression, and is being used more widely as more recent browsers give it better support.

Is this an area where a proprietary standard might lead to big $$$?

Probably not, since PNG fills that niche.

Re:Admissibility of Digital photographs as evidenc

[e40]

That article is pretty pathetic. It doesn't even mention this, which can be used to verify that an image is original and unmodified.

Reads like vaporware

[fnurb]

Neat turn of phrase, but the implication that those statistics and regularities that differentiate a "natural" photograph from a "doctored" one (which is not a "random" distribution of pixels, BTW) are well-understood, at least by the claimant, is as questionable as the conclusion from the above statement that the essence of intelligence, that differentiates a Shakespeare from a monkey, is well-understood.

Re:What kind of digitized photos does this work on

[GlassHeart]

Exactly. You can view the JPEG compression process as mapping a large potential set of originals to one single JPEG image. However, if you can prove that you can generate the exhaustive set of originals (no other original can result in the JPEG in question), and that none of the originals pass the authenticity test, then you've proven that the JPEG could not have been genuine.

Reads like vaporware

[fnurb]

"Natural digital photographs aren't random," he says. "In the same way that placing a monkey in front of a typewriter is unlikely to produce a play by Shakespeare, a random set of pixels thrown on a page is unlikely to yield a natural image. It means that there are underlying statistics and regularities in naturally occurring images."

Neat turn of phrase, but the implication that those statistics and regularities that differentiate a "natural" photograph from a "doctored" one (which is not a "random" distribution of pixels, BTW) are well-understood, at least by the claimant, is as questionable as the conclusion from the above statement that the essence of intelligence, that differentiates a Shakespeare from a monkey, is well-understood.

Re:What kind of digitized photos does this work on

[mysticgoat]

You make some good points.

To me this strikes me as the same sort of "solution" as DRM is, sure it stops Joe Six-Pack from putting Britney Spears' head onto a porn stars body, but it will not stop anyone who knows what they are doing when it comes to digital image manipulation.

Another point is that this technique is going to have limited usefulness since some degree of image manipulation is a normal part of publishing photos-- whether on the web or in other media. Things like sharpness, color balance, and resolution are tweaked to fit the presentation. It seems to me that any authentication technique sensitive enough to identify good forgeries would also declare every image published by the NYT as being faked. I don't see how that would be useful.

There are circumstances where it would be good to be able to authenticate raw, unedited photos. But they are pretty few in my life.

Most of the time my concern is not going to be about whether an image has been manipulated-- I expect professionals and better amateurs to do some manipulation-- but about the degree and nature of the manipulation. Did the artist do something more than remove the outhouse that was in the background and clean up the dark shadows? That kind of question can best be answered by looking at the reputation of the artist rather than the image he's produced.

Fake Videos: Robot built from a Mini Cooper

[antispam_ben]

This link: http://www.r50rd.co.uk/research/internal/v2i/engin / was floating around a few newsgroups last year, showing videos of a robot with fairly amazing capabilities. Here's the Usenet discussion (comp.robotics.misc, "Robot built from a Mini Cooper?") where I first saw it:

http://groups.google.com/groups?hl=en&lr=&ie=UTF-8 &c2coff=1&threadm=14f6fe8.0403090913.81e3f13%40pos ting.google.com&rnum=1&prev=/groups%3Fq%3Dmini%2Bc ooper%2B%2522Is%2Bthis%2Bfor%2Breal%2522%26hl%3Den %26lr%3D%26ie%3DUTF-8%26selm%3D14f6fe8.0403090913. 81e3f13%2540posting.google.com%26rnum%3D1

As I (posting as "Ben Bradley" [should I go ahead and put my real name in my profile now?]) say in my posts, the videos appeared quite impressive at first viewing, but were easy enough to see the problems and deconstruct upon repeated viewings with a critical eye.

A search of earlier stories shows Slashdotters spotted the video's fakeries as well: http://slashdot.org/article.pl?sid=04/03/09/222227 &tid=159

The point being that these algorithms for detecting fakes may indeed work well, but I don't see them as being a monumental thing that "We Are Now Able To Detect Faked Photographs" but rather an incremental thing, that uses a computer to verify what we can already find out with a little probing and [knowledgable!] common sense.

Will Lead to Better Smoothing/Feathering

[BrianMarshall]

If you add something to an image, or take something out by replacing a region of the image with something else, you get a boundary. If the boundary is a line ("cut and paste"), it can be visible to the eye and statistical approaches could detect it.

So you use smoothing and feathering to turn the boundary into a region. This can make the change harder to detect by eye, but the boundary would presumably look different than parts of the original image, statistically, (the changes from one pixel to the next) - a region that has been smoothed will be smoother.

Presumably, this will lead to better smoothing and feathering techniques, maybe using (more) pseudo-randomness - slight additional random changes from one pixel to the next, but randomness that has the same statistical properties as adjacent parts of the original image.

Re:What kind of digitized photos does this work on

[Anonymous Coward]

Exactly. At the photo-lab I used to work at(in the digital department, where we did almost all of the photo-restoration/retouching jobs),not only could we print to transparency or negative, they had(have) laser printers that directly develop photo paper in the machine. I still use their services for prints of my digital artwork, as the color gamut is considerably bigger than inks, not to mention remaining in one color space from conception to print.

I wonder how some of todays movies would fool "the test".

Useless waste of taxpayer's money

[Randym]

With money from the Air Force, Professor Fridrich is designing a camera that takes two pictures at once: one though the camera lens, and a smaller one of the photographer's iris. The iris image along with the time and place of the photo session would then be compressed, encrypted and instantly hidden within the larger picture just taken.

Does the good professor seriously think that people -- especially people who are aware that their camera utilizes this technology -- *only* take pictures by putting their eye to the camera when they click? Ha! If you use a stand, your 'iris' picture could be *whatever you wanted*. I know my 'iris' picture would be a picture of an iris, all right -- but not one in any database. Maybe I'll take a TIFF of my *cat's* iris, photoshop recolor it in, oh, *lavender*, print it on nice photo paper, light it right, and hang it an appropriate distance from the 'iris' lens -- that'll freak 'em out, *for sure*. 8^D

BTW, I don't know about other states, but this kind of steganography is illegal in Michigan under the new Super-DCMA law.

That's the great thing about government research -- no idea is *too* wacko .

Re:Useless waste of taxpayer's money

[harlows_monkeys]

Does the good professor seriously think that people -- especially people who are aware that their camera utilizes this technology -- *only* take pictures by putting their eye to the camera when they click? Ha!

You've got it backwards. The situation where this would be useful would be where you are taking a picture of something, and you want to protect against someone else altering it and claiming that the altered picture is the one you took, or you want to otherwise prove at some future time that you took it. In this situation, you'd want to make sure that the camera sees your iris.

For example, consider a cop taking a photo of a crime scene. Later, at a trial, proving that the cop took the photo could become an issue. If the cop made sure that it was stamped with his iris information, that would be very useful.

Re:This camera's authentication feature *is* hacka

[rgmoore]

This can be easily defeated by hacking the 1Ds's firmware. The private key and algorithm used for signing are most likely encoded within the firmware (easy to understand - this camera uses an embedded PowerPC control processor), rather than on a specialized crypto chip. Just pull out the algorithm and key, resign your CRW files and they're authentic.

It looks as though the function isn't (completely) built into the camera itself; it requires an external card to do some part of the signing. Your fundamental objection stands, though. As long as the system doing the signing is in the hands of the would-be forgers, it is vulnerable to being hacked.

I think that what this really shows is that signing has limits. You can create a perfect, tamper-proof signing system for your camera, but it can only prove that the image hasn't changed after leaving the camera. The system can't tell if the image coming through the lens is genuine. It will sign a picture of a printed version of a manipulated image as readily as it will sign the original, unmodified version.

Commercial Software

[JThundley]

"Commercially available software makes it easy to alter digital photos,..."

Oh, so you're saying that The Gimp isn't user friendly? Flame on, motherfucker!

Follow the link...

[Anonymous Coward]

Click on papers, then you can find at least 2 papers dealing with the same thing. And these papers are available.

This is typical for some (especially US) researchers, to rehash the same idea over N+1 papers just to get more publications. I wish people would stop doing that! Part of blame belongs to journals, you want all the prestigious ones to carry your clone-paper.

Different goal

[Anonymous Coward]

Actually the main goal is neither detecting forgery nor to produce it. According to their page http://www.ists.dartmouth.edu/ISTS/research_progra ms.htm#dhdi the main goal is to detect steganography. Thus the forgery detection, and its potential application to validate photographic evidence, may be only a pleasant byproduct of the research.

On the other hand, all the Farid's technical papers mentioned on his page (linked several posts above) seam to deal with the forgery detection, the detection of steganography being targetted only in several press releases. May be the forgery detection is the first step, the only to be actually taken yet. Or may be the algorithms targetting directly the main goal are kept secret.

What?!

[bruthasj]

My tinfoil hat just came in from Amazon. Ask Slashdot: Should I return it? Is that question Googleable?

Thanks

Worth1000?

[RevRagnarok]

I'd love to see the data run on W1K submissions...

Re:Admissibility of Digital photographs as evidenc

[evilviper]

Bullshit!

The story is about a fingerprint image that everyone knew was enhanced to make a better match. This new method would do NOTHING AT ALL in that case.

Now, to the subject you brought up. The fact of the matter is, digital images are no more easily faked than film images. Sure, you can mess around in GIMP and make something that would fool your family members, but it's not going to fool police experts.

Photographic evidences has a lot of tell-tale signs when it has been altered. Whether film or digital, visual inspection by an expert can conclusively tell if something has been manipulated. Besides, if digital is so untrusted, why not just digitally modify an image, and then transfer it to film, and claim you took it with a film camera.

That's not even beginning to discuss the additional information that digital has along with it, that can help further verify it's authenticity. Any expert can tell the difference between a JPEG saved by a digital camera, and a JPEG saved by Photoshop. They don't use the same equations, so the output is different, and there are tell-tale-signs in the byte sequence (no I'm not talking about the comment inserted by Photoshop/GIMP, although that would be good for easily catching amatures).

Plus, things like time-stamps on the computer you transfered the image to can be verified. You can tell if the clock was changed, you can tell if there are other version of the file (now deleted) on the hard drive. You can see if the image on the CF card was created by the camera, or sent from a computer. etc.

Some forgeries are SO lame

[AstroSurf]

IIRC, that issue of Whole Earth that showed the UFOs over San Francisco was particularly notable because the UFOs had no shadows. Not too convincing. :/

CGI

[Anonymous Coward]

Would this algorithm be able to detect expertly-rendered CGI scenes? There is more to digital fakery than mere Photoshopping of existing images; fake pics can be manufactured wholesale.

An example: using matchmoving software like Realviz or Boujou, I can swap the license plate seen on a car in a handheld video, then run the result through to an old VHS camcorder and pin someone to a crime scene, who was never there.

I'm still waiting for someone to pull that off, because the consequences for evidence in courts are just huge.

This is Pretty Cool

[stevemm81]

Now I can recognize all those pictures of myself that don't exist...

Sig semper tyrannis.

Re:What kind of digitized photos does this work on

[flonker]

The exhaustive set of originals is going to be on the order of 2^(8*(bytes_in_lossless_jpeg - bytes_in_lossy_jpeg)). This is going to be many orders of magnitude greater than the number of electrons in the universe times the number of measurable cycles of time. Meaning, the only way it can possibly be done is with quantum computers or some other revolutionary advance in computing.

The Porn-A-Kid Brief: Is That Child-Porn Digital?

[nusratt]

"Perhaps I am misreading your intent"
Yes, but it's still my fault, and this time I see more clearly why:
it's because I never made it clear that I'm not thinking about the specifics of Farid.

FORGET Farid, FORGET about the particulars of JPEG versus other formats, etc.
I'm talking about the more general, abstract, case.

First, let's dispose of the "perfect knowledge" case, because that's easier.
Assume that "the world" has an "undefeatable" way to detect image alteration.
Yes, I acknowledge that the very word seems an oxymoron -- how can you prove a negative, how can you prove that someone hasn't found away to defeat detection?
But it's irrelevant: I think we can agree that, if "undefeatable" detection WERE available, then no possessor of kid-porn could succeed in falsely claiming that his photos are merely digital fakes.

Now let's take the other case, of detection technology which exists but is imperfect, i.e. defeatable. (btw, for the purposes of this illustration, known-defeatable detection is really no different than the case of the "unproveably perfect" detection technology.)
Let's revisit the criminal-trial scenario . . .

-- Mr.X is arrested for having child porn images.
-- As I mentioned before, a U.S. court recently ruled that child-porn doesn't violate the Child Online Protection Act, if the porn was actually produced without the involvement of real children.
-- Mr. X's defense: "The images are digitally altered, no actual children were involved."
-- Prosecutor: "We had the images analyzed by the world's foremost scientific expert, who attests that these images show no sign of digital alteration. Therefore, your claim of digital alteration is not credible, and I believe the images to genuinely portray children, and so should the jury."
-- Mr X: "Yes, I know your expert's detection techniques; and so, after creating the altered images, I finished the job by using special techniques designed specifically to defeat those detection techniques."
-- Prosecutor: "WHY DID YOU GO TO THAT EXTRA TROUBLE?"

Now, put yourself in the defendant's shoes.
How can you answer this last question from the prosecutor, in a way which sounds plausible?
If you can't produce such an answer, the jury is now more likely to believe that the prosecutor is right, that the images portray real children.
And even if you can produce a plausible-sounding answer, you might then be compelled to *demonstrate* that you're actually capable of doing what you claim to have done (including your claimed counter-detection capability.)

THIS is the novel legal consequence of work LIKE Farid's (in my opinion).
The legal landscape has been changed for these kinds of cases, EVEN if the detection technology is imperfect.
Defense arguments like Mr. X's haven't been made impossible, but they're a lot more difficult.

Re: "you can get away with it?"

[nusratt]

"So if the [illegal kid-porn] image is real, you can get away with it by doing a little image manipulation to [make the digital-alteration detection-test conclude that the image isn't genuine]?"

Basically, yes. See The Porn-A-Kid Brief.

Re:What kind of digitized photos does this work on

[GlassHeart]

JPEG works on 16x16 pixel blocks at a time. This means that you generate 16x16 tiles and see if they can compress to the same portion of the JPEG image in question. In other words, for every 16x16 block you have to generate 16*16*(2^24) distinct images and compress them. That's about 4 billion (2^32) distinct images, which should be roughly equivalent to compressing 1.4 million 1024x768 JPEG images (3072 blocks per image). With 30 fps real time MJPEG compressors, that should take about 13 hours to compress.

To prove that the evidence was not doctored, a full 1024x768 image would take about 1,664 days (about 4.7 years) to complete at 30 fps. The good news is that the task is entirely parallelizable, so fifty such encoders would find the complete set of originals in just over a month.

So no, nothing astronomical as far as I can see.

Re:The Porn-A-Kid Brief: Is That Child-Porn Digita

[Alsee]

Why are you assuming that the person is question is the person who created them in the first place?

If we accept that possession of fiction is not criminal, then downloading and possession fiction is not criminal. We can now consider two possibilities (1) some person not involved in the case created the fictional images. Maybe that person made special effort to eliminate signs of tampering, or maybe the methods used just didn't happen to leave anomolous traces. Either way you'd have to find the person who made the image and ask *him* about that, not the person merely in possession. (2) Maybe they really are real images, but they were falsely labeled as fiction. In that case the person in possession would not be guilty - lacking mens rea, criminal intent. If you pick up a box labeled plastic toy handgrenades, and someone hands you a box of REAL handgrenades, you are not a criminal.

Criminalizing possession of fiction is just silly, and the problems I listed above merely serve to highlight a broader problem - the idea of "criminal possession of information". I find "criminal possession of information" a rather disturbing concept.

We have perfectly good laws making it criminal to abuse a child. We have perfectly good laws making it criminal to intend to cause a crime to occur (including the abuse of a child). We have perfectly good laws making it criminal to contribute or aid someone else to commit a crime (including the abuse of a child). Any bastard guilty of any of those crimes should be thrown in prison.

If someone committed any of the crimes listed in the last paragraph, then fine, we're already throwing them in prison for that. However if the only charge ammounts to "criminal possession of information" then by definition we are talking about somone who has *not* abused a child, did *not* intend to cause a child to be abused, and did *not* contribute to the abuse of a child. If you want to argue my last sentence is wrong, that we are talking about an actual criminal doing actual harm, then I say you are actually talking about somone who is actually guilty under one of the laws in the previous paragraph and not someone guilty of only "criminal possession of information".

We certainly don't imprison someone for possession of an image of a bank robbery or other crime, even if that image was created by the criminal himself. And even if we did, it would be just silly to expect anyone in possession of a copy of essentially any movie or TV show to prove that it is actually fiction.

-

Re:So much for my faked photo of Nixon and Elvis

[Richard M. Nixon]

I was coming up with concepts for an album cover, and this is one of the ideas that I came up with.

I hope you didn't use it, because I never received any royalty checks.