Interpol Unscrambles Doctored Photo In Manhunt

jackpot777 writes in with an AP story out of Paris reporting that Interpol has distributed photos of a man suspected of sexually exploiting children. The images were recovered from pictures taken off the Internet in which the man's face had been blurred using something like Photoshop's Filter > Distort > Twirl tool. German police were able to recover recognizable images of the man, whose identity and nationality are not known. Interpol would not discuss the techniques used to recover the images. jackpot777 writes: "It does show one interesting facet of internet privacy that has also been noted with topics ranging from reading blurred check numbers in images to Google's plan to blur out license plate and face data for Street View. And that is: blurring is not the same as completely obscuring. As computers become more adept at extrapolating data of different types, your identity isn't safe unless you completely cover all those identifying features."

Pictures

The pictures [interpol.int] can be seen on Interpol's site.

Re:Pictures

At least in this case, it's dealing with filters applied by software, so the algorithm can be examined, and it's perhaps reversible, whilst in 24 they often apply it to things like poor quality or low resolution cameras, and magically enhance the details.

I've got a friend that was charged for burglary one time and the company that charged him submitted their surveillance video footage to some supposed forensics team so they could see if they could derive his face from the blurry video. What was brought to light was that the idiots also submitted my friend's work ID, and an old one that didn't really look like him any more, with the video. The team then returned a video that showed how they "matched" my friend to the person seen in the video...they morphed several stills grabbed from the video with....you guessed it...my friend's ID and THEN they showed their derived picture right next to the old ID. I took one look at it and told him they had absolutely zero case against him if that's all the evidence they had. I didn't even have to show up as an expert witness since the judge was wise enough to realize what was going on.

Re:Pictures

Well, I'd guess the twirl is a convolution filter followed by a twist. If you can separate it out then you can reverse the twist, and then deconvolve the resulting blurred image, you get the original image. I'd guess. But it's been a while since I did computer vision, and it's probably more complex. :P

Re:Pictures

A convolution filter would allow you to sharpen or blur the image, or maybe even pick out high frequency detail.

The spiral distort effect is simply a mathematical function to map one point in a rectangular mesh to another. You basically convert integer pixel coordinates into a floating-point coordinate system with the origin at the centre, apply a rotation based on the distance from the origin, convert back into integer pixel coordinates and transfer the pixel data.

Consequently, since every pixel is remapped to a new position, the transformation can be reversed.

Way easier than that....

Edit >> Undo

Re:Pictures

Thoughts on why they want to keep the technique secret:

"Woah, we caught a break, sir. That pedophile just used a twirl filter."

"No kidding? Did you get his face out of th..."

"Already done. Hopefully more of those assholes will use that twirl filter."

"Yeah, good point. Keep it quiet. I'll make up some story about secret techniques and taking six months."

"Oh, sir, one more thing. We really should get a legitimate copy of Photoshop."

Re:Crucial overlooked ideas

They could then look at the twirled test image and come up with a mapping of twirled pixels to untwirled pixels. This information could be used to "untwirl" the original image by grabbing the pixels at the twirled coordinates and moving them back to where the mapping says they probably originated.

It probably helps a LOT that in several of the images, there's a strong line visible in the background. Measure the twirl of that, you've got your benchmark right there. Center of the twirl is probably easy enough to locate too. So there's your twist, and where to apply it.

It's a good thing so many criminals are dumb. It's the smart ones that you have to worry more about.

Re:Crucial overlooked ideas

It's a good thing so many criminals are dumb.

Who says he's dumb? If he carefully photoshopped someone else's face onto his, and then applied the easy-to-remove swirl, he now has the entire planet searching for the wrong guy....

Amazing technology

Amazing, indeed [imageshack.us]

Re:Amazing technology

It doesn't work that well. They guy on the left looks all wonky and strange.

Links to the photos

Let's get this guy behind bars by the end of the week. Photo 1 [waffleimages.com], Photo 2 [waffleimages.com], and Photo 3 [waffleimages.com]. Photo 4 [waffleimages.com], Photo 5 [waffleimages.com], Photo 6 [waffleimages.com], Photo 7 [waffleimages.com], Photo 8 [waffleimages.com]

Interpol not the ones to descramble

From the interpol web page it says:

These pictures have been produced by specialists from Germany's federal police force, the Bundeskriminalamt, working from originals found on the Internet, which had been digitally altered to disguise the man's face.

Re:Interpol not the ones to descramble

Ummm, about your sig. In this context? You're not him, are you?

The actual AP article

.. can be read here [google.com].

Super-sekr1t unblurring techniques

Surely Interpol's top-secret image-unblurring technology is just a matter of applying the Twirl effect in the opposite direction at the same location, and perhaps applying some image-enhancement plug-ins to the resulting area? I doubt it's anything one couldn't do with off-the-shelf software.

Re:Super-sekr1t unblurring techniques

The AP article did mention that AP were able to produce an almost recognizable image using commercially available photo editing software but not a good as the one Interpol had produced.

Wild ass guess ahead...

Interpol geeks probably ran some tests to determine approximately how much twirl was applied to the original image and then created a 24bit image slightly larger than the twirled area assigning a unique 24 bit value to each pixel and then applied the same amount of twirl.

They could then look at the twirled test image and come up with a mapping of twirled pixels to untwirled pixels. This information could be used to "untwirl" the original image by grabbing the pixels at the twirled coordinates and moving them back to where the mapping says they probably originated.

Of course there would be some pixels lost and extra pixels created during the original twirling but chances are the original image could be approximated fairly well by interpolating between the recovered pixels. You'd not get a picture perfect result but something somewhat blurry as can be seen in the recovered pixels.

Of course they might have done something more mathematical but if I was going to try this myself I'd probably just give the method I described above a shot first and see if I came up with something looking like a face.

Re:Super-sekr1t unblurring techniques

Yep, I don't see what other way there is to do it rather than just reverse the process :P It's funny seeing some people say it's 'amazing', and also it's a dumbass way to try to disguise something. I don't think the summary is accurate in thinking that blurring can be undone to the same extent as to get a license plate back or whatever.. it would be possible to get a bit of detail back if there wasn't too much blurring being done, but say you applied a very heavy gaussian blur to a section of a picture, everything would basically get mashed into one colour..

Re:Super-sekr1t unblurring techniques

That's why when I need to obscure a face (when working on images for a medical journal) I use the mosaic filter after running a heavy gausian blur --- leaves something recognizable as a face, but w/ too little information to reconstruct even a postage stamp (~10 x 16 pixels).

William

Re:Super-sekr1t unblurring techniques

Another fun technique is to just paste something plausible there—like another face, or pieces of other faces—before blurring. I often do the same thing when blurring out numbers. You might be able to get away with a plasma fractal with appropriate skin tones.

I figure it gives the hacks something to get excited about until they realize it really is gibberish. :-)

--Joe

Re:Super-sekr1t unblurring techniques

Your method is not entirely foolproof. The problem is that someone can repeat the steps you took and compare the results. They could photograph a person in the same pose, apply the same or very similar filters and if the results match, there is a good chance the source images were also quite similar.

A similar technique was used to guess blurred out numbers on cheques, passports, car number plates etc. Simply run through all possible combinations of letters and numbers, applying a mosaic each time until the mosaics match.

Re:Super-sekr1t unblurring techniques

If you really want to obscure something in a non-reversible way, remove the data from the image (overwrite it with a black box or something).

if you are from a government, remove the data from an image with the alpha channel and don't uncheck "save color values from transparent pixels" (in gimp).

Re:Super-sekr1t unblurring techniques

It's easier to just draw black boxes over it then save as XCF (since if you try to save in another format it warns you about "flattening" the image, whatever that means)

Re:Super-sekr1t unblurring techniques

I've just tried this using Photoshop's twirl plugin, and with a little tinkering arout I could get a fairly good descrambled picture in only 10 minutes.

With more time and higher quality images, I'm sure it wouldn't be any trouble at all, it just needed the initial insight to use the "swirl in opposite direction" idea.

Re:Super-sekr1t unblurring techniques

Second. In 30 seconds I took a picture of my wife, did a maximum twirl of 999 deg.
Then I did a -999 deg twirl, and it totally undid it - almost perfectly.

I'm surprised it worked so well actually, I was assuming that it would still be more distortored since I thought the result would be more of a hash, but the algorithm used looks to be 99% reversable - at least in the case of photoshops' distort/twirl effect.

Re:Super-sekr1t unblurring techniques

Show us pictures of your wife!

P.S. Wow, this comment is on-topic for once!

Re:Super-sekr1t unblurring techniques

Amazing! I tried it too. Did the twirl thing as far as it would go then used "undo" to reverse it and you really could not spot any difference AT ALL!

Rellying on the CSI effect

Yes, for the average /.er this is absolutely trivial :
the idiot used a filter that just moved the data around in predictable way (in circles), and the police did transpose the data in the opposite direction and got the picture back. The picture was not blurred at all (in the mathematical sense of lowering the resolution).

Interpol bragged about it not because of some obscure technical feat. They bragged about as a PR stunt, in order to take advantage of the " CSI effect [wikipedia.org] ".

Joe 6 pack, has recently started to understand that incredibly big zooms, with some magical "picture enhancement effects" that keeps incredible amount of details - as done by Deckart in Blade Runner, or regularly featured on CSI - can't be actually achieved in real life. Because everyone is criticizing those shows for the lack of realism in their zooming achievement.

But now Interpol pulls this PR stunt, where they show how they managed to recover the identity of the maniac. Now people every where are starting to think "Oh may god ! They actually have the technology ! They can "enhance" pictures and get the faces back !". The goal of Interpol was to instill fear in would-be criminal who would hope to stay anonymous with some photoshop tricks tricks. Maybe this wasn't the only stuff that was openly criticized in CSI but that was secretly doable by the real police. Now cue-in some armchair conspiracy theorists, who could pretend that the whole criticizing of "unrealistic police TV-shows" was a government conspiracy to cover up technology that actually exist (additional points earned if technology is of alien origin), or they could say that government has put a backdoor inside Photoshop that does keep the blurred faces saved in steganography (bonus point for using buzzword).

They are creating a climate of FUD, in the hope to deter would-be criminals.

Blurring different from twirling...

A twirl is essentially shifting pixels around an image, and is designed to keep as much information as possible.

A blur on the other hand, especially a gaussian blur, will mix pixels together in such a way that any recovered image will be one of many possible outcomes.

Then again, removing information, by pixellating for example, would be best.

Re:Blurring different from twirling...

Surely the most effective filter would be a GIMP mask?

Re:Blurring different from twirling...

Pixellation is only effective if done really, really coarsely ... so you might as well just opt for the classic black bar to begin with.

The link about cheques in the summary tells more (if it's that old article I think it is).

Re:Blurring different from twirling...

Or do it in hardware instead: Black bar glasses [stupidiotic.com].

Re:Blurring different from twirling...

Who are the alleged victims?

"Alleged" implies that a picture of an adult molesting a child, constitutes only an "alleged" crime.
"There were no rape report from anyone" implies that a picture of the act being conducted isn't enough. That the 4 year old Cambodian sex slave needs to file a proper report.

In a discussion regarding the photographically documented molestation of small children, you want to expand the discussion to include statutory rape allegations between teenagers. How about we also talk about taxidermy and monster trucks? Because if you think there's any similarity between the teenage sex issue and child molestation, then any "discussion" with you might as well go down those tangents as well.

Hardly Rocket Science

>Interpol would not discuss the techniques
I showed this to my PS using friend and he shurgged, said 'Just do a radial blur in the opposite direction' and 30 seconds later had a picture about 80-90% as good as the one they're waving about as being the result of some super secret methodology.
It does strike me as a bit stupid explaining it all - now crims will just use better techniques for blurring themselves out. The media, law enforcement agencies are doing this more and more and it's insane - "we just had an idea for a terrorist attack that might happen and here it is in full", "This is foresnic evidence that allowed us to catch the crim" and so on.

Re:Hardly Rocket Science

I showed this to my PS using friend and he shurgged, said 'Just do a radial blur in the opposite direction' and 30 seconds later had a picture about 80-90% as good as the one they're waving about as being the result of some super secret methodology.
It does strike me as a bit stupid explaining it all - now crims will just use better techniques for blurring themselves out. The media, law enforcement agencies are doing this more and more and it's insane - "we just had an idea for a terrorist attack that might happen and here it is in full", "This is foresnic evidence that allowed us to catch the crim" and so on.

Yup, they spun it (pun intended) into cheap PR. The problem is, it's not that they are super smart, it's that the criminal was super stupid.

And it'll make anyone with basic image processing skills question their overall expertise if they'd brag about untwirl.

That said, the average folk will definitely be impressed. I knew a guy who inverted his photo in attempt to protect his identity (no, he didn't molest children). Imagine his shock when I took the inverted photo, inverted it again arriving at the original.

To him I'm probably some sorta super genius who used sophisticated data restoration hack. To a guy with basic knowledge, it's nothing worth noting.

To see how blur can restore detail not visible to the naked eye, check out Focus Magic [focusmagic.com]. Not as easy as untwirl, but gives you an idea. This is because the blur distribution (usually gaussian if digital, or linear with cameras) gives away the possible origin position of the pixels.

If you pixelize however, with big enough square, you lose real resolution and that's much harder to restore anything interesting out of (it's not like in movies, with the unlimited extrapolation techniques, as we all know).

Other gotchas: covering with black rectangle but leaving it only 1-2% transparent. Looks solid, but data can be recovered.

And a very common other method: people keep leaving their name and camera model in the meta info of the image. Easy to check out via right-click>Properties in Windows.

PS: it was "twirl", not "radial blur" btw.

Re:Hardly Rocket Science

I also wonder if you could recover an image of someone's face from pixelated video. If the camera or the person is moving but only slightly then you may be able to determine the x,y movement of the whole image from the non-pixelated parts of the image. From this you can then consider each of the large pixelated pixels as a sample point on the person's face, and as they move you aquire additional sample points. Over enough time, say a 5 minute interview, you might be able to reconstruct a recognisable face.

Ah, good point indeed. If it's a video, yes, you can restore extra detail.

There are lots of cameras out there which use a simple version of this trick to shoot higher res photos than their matrix is (by shooting several photos with sliightly offset matrix and assembling those).

And there's already software in wide use which can take existing video footage of, say, recording a page of a book for a while with low resolution, and using the minor motion/shifts in the frame to automatically arrive at a much higher (and accurate!) resolution image. It's amazing the amount of detail it can restore.

Since pixelization is in fact reducing the resolution, the same applies there.

I guess the only sure method is not to leave anything that can be analyzed. Don't wanna be recognized.. ? Don't allow to be recorded/shot.

a better solution

As computers become more adept at extrapolating data of different types, your identity isn't safe unless you completely cover all those identifying features.

Yes, or you could just stop molesting children and photographing it.

blurring != obscuring; true, but...

true, blurring isn't the same as obscuring. That said, a twirl/swirl filter isn't a blur filter either. A twirl/swirl filter relocates pixels from position A to position B. The original pixels are still largely there, you just have to move them back from B to A. That's what Interpol did here - kudos to them for figuring that out. But a blur filter doesn't just relocate pixels - it blends a bunch together. Now don't get me wrong - there's certainly deconvolution methods to reduce blur - especially motion blur - ( one example software: http://www.focusmagic.com/ [focusmagic.com] ) but you're not going to be able to just take any heavily non-motion blurred image and get a supersharp result back. Other techniques, such as pixelization, are even worse to restore - you may as well not try.

-That- having been said.. yes, obscuring does tend to be better.. as long as it's a proper obscuring and not some half-hearted attempt by a news station where an interviewer / whatever has said to want to be inrecognizable, and then you just get a dark silhouette of the person where you can 1. still make out the silhouette, 2. their voice goes unaltered, 3. bump up the brightness enough and you can even make out a face or, in the case of yea olde license plate, a black bar that is supposed to 'track' the license plate properly, but the person applying the bar is a lazy-ass tracker and it 'swims' over the plate, revealing tiny bits of the bottom/top on certain frames - not too much guesswork involved to figure out the proper license plate, as even with multiple possibilities, only one is likely to match the type/color of the car when looked up on the interwebs.

Now then... Let the "what if somebody photoshopped somebody else's head on there first, then applied the filter, now some poor innocent sap is framed!" replies begin.

How to remove numbers/faces from a picture

Just cut that part out of the image, and use some algorithm to fill in the blank space without depending on the original data in that area. Hey, if you are blocking out a license plate in that way, the algorithm could even fill in something that looks like a license plate but with some random characters instead of the original ones. It doesn't matter what the algorithm does, since the information you wanted to hide was completely removed before using that algorithm.

Once the data's gone, it's gone...

As computers become more adept at extrapolating data of different types, your identity isn't safe unless you completely cover all those identifying features.

Uhm, no. As other posters have pointed out, all they did was reverse the distortion applied to the image - which in this case didn't really lose much information, just nudge it about. If you blur out someone's face, the detail can never be recovered. No, not even by the NSA. The information is lost. You *can* sharpen up edges and improve contrast, but if the information just plain isn't there any more there's not a lot you can do.

Think about it this way. A digital image is just a string of numbers. If I take a string of numbers and apply a "filter" to it then I get (0.4, 3.0, 6.2, 3.4, 5.4, 5.8, 2.6). From that, can you work out what the original values were? Possibly, because my filter is very simple. However, you don't know how much precision has been lost, or what the initial values were, so it would be nigh-on impossible to work out the original values.

Incidentally if anyone does work out the original sequence, I'd love to hear about it.

Re:Once the data's gone, it's gone...

Also, for check numbers, as well as facial features, we expect a quite specific structure. This means that the set of possible original images is not uniform, so it's quite possible to detect a remarkable level of detail (like the specific check number, if the font is well-defined), even when a lot of information has been lost.