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."

Comments

Pictures

[avij]

The pictures can be seen on Interpol's site.

Re:Pictures

[petermgreen]

almost certainly

the way I would approach this is to get hold of all major photo editing packages and then do a twirl in a known location on test images. working from those twirled test images it shouldn't be too hard to work out what is being moved to where. If the packages are scriptable this is a lot easier. If not then its a lot of grunt work.

once you can reverse a twirl in a known location it then just becomes a matter of moving your twirl reversal tool arround until you get a sane looking image.

Re:Pictures

[fractoid]

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

[allcar]

When I read the headline, I was intrigued. It makes the whole process sound like something from 24. The sort of thing that normally gets me shouting "That's bloody impossible!" at the TV and annoying my wife. However when you look at the actual photo, you realise that the guy, as well as being evil, is a complete fuckwit. At first glance, you know that this is a reversible transformation. This guy should be given a Darwin Award for utter stupidity.

Re:Pictures

[neodiablo22]

This was covered during an interview on 60 minutes last Sunday. http://www.cbsnews.com/sections/i_video/main500251.shtml?id=3340375n

Re:Pictures

[mdwh2]

When I read the headline, I was intrigued. It makes the whole process sound like something from 24. The sort of thing that normally gets me shouting "That's bloody impossible!" at the TV and annoying my wife.

Me too (well, minus the wife :p) 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 liked X-Files for this though, where they would enhance a single pixel of someone's nose, to reveal there is in fact a ghostly alien hidden up there.

Re:Pictures

[mdwh2]

The most trivially simple explanation would be that child abusers are as likely to be female as male.

Indeed - and anyway, if that isn't true, we instead have the statistic that "men are much more likely to be abusers than women". But I don't see most men being worried about that, nor do I see it being used as a tool of prejudice against men.

Re:Pictures

[insertwackynamehere]

Well I tried this out in the gimp with the Whirl and Pinch feature..whirling can be reveresed, his mistake was that he didn't pinch the whirl. When I pinched the whirl I could no longer de-whirl the image by just doing a negative of the original whirl angle.

Re:Pictures

[mikael]

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.

Re:Pictures

[Threni]

> This guy should be given a Darwin Award for utter stupidity.

In which way does he qualify for a Darwin Award? Clue: There's more to one than being stupid.

Re:Pictures

[fishbowl]

>In which way does he qualify for a Darwin Award?

The appropriate punishment for this crime is castration?

Re:Pictures

[ehrichweiss]

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

[Anonymous Coward]

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

It's not done that way, because that would result in strong aliasing. Filters work the other way around: For each pixel in the target image, calculate the area in the source image which is mapped to that pixel, or more likely just the center of the area because that's a lot faster. Then you can resample the source image at that position and thereby avoid aliasing. If the position isn't an integer coordinate, interpolate between the surrounding pixel values. That's the step where a small amount of convolution occurs, depending on the interpolation that is used. If you apply a twirl effect and then the inverse twirl, you'll notice that a slight radial blur remains. That's the convolution.

Re:Pictures

[diggum]

Way easiier than that, at least with the photos the news org were showing. Make a circle selection, taking care to line the edges up with where the twirl circumfremce is, and use photoshops twirl filter to go the other way. We did it in the office and came close to the image interpol released, accounting for the fact we had a very low res source. (and i can assure you that we were not interested in anything higher resolution. That guy is creepy as hell, and that shirt was so ugly.)

Re:Pictures

[AndyG314]

I don't find this shocking that someone was able to reverse the blure algorthum of a popular software. It is doubtfull that blur (or any other photoshop filter) was designed to be a one way function. If you know what was used, do it backwards and get the origional. Also this shouldn't raise much privacy concern, Use a black rectangle instead of the blur and your all set. That would be much more difficult (impossible?) to undo.

Re:Pictures

[allcar]

In which way does he qualify for a Darwin Award? Clue: There's more to one than being stupid. He's committed a hideous crime, then posted a picture on the internet that can be easily unscrambled. This will dramatically improve his chances of being caught. When caught, he will be locked up for a long time. In all probability, his fellow inmates will then cut his bollocks off and make him eat them. I rather think that removes him from the gene pool, for which he deserves a Darwin award. Having said that, a homosexual paedophile has probably already opted out of the gene pool.

Re:Pictures

[badasscat]

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.

And the article summary here definitely confuses the two, talking at the end about blurring license plates as if it's the same thing as what this guy did.

I have yet to see any evidence that a properly blurred image can ever be recovered. People keep talking about it, but they're talking about getting blood from a stone. Blurring is by definition a loss of detail. You cannot restore lost detail; you can only try to approximate what that detail was. (The article posted here a while back, and linked through the summary above, is talking specifically about extracting numbers from a mosaic'd image, not a blurred one. In any case the author says specifically that it only works when you're trying to choose from one possibility out of ten choices, ie. a number.)

Let's say you have an image that you blur to the point where it's nothing but a sheet of grey color. There is nothing in the world that could ever revert that image back.

The only question is how far you have to blur something to get to that point. Some people may not go far enough, and their images can be extracted. But there is a point of no return.

What this guy did, though, was use a filter that preserves nearly all detail but simply distorts it. And any distortion can be un-distorted. That's a different thing entirely than extracting detail where there is none to extract.

Some types of image manipulation destroy data; others don't. "Blur" destroys data. "Twirl" just moves it around. That's the difference.

Re:Pictures

[debuglife]

Thats a very good explanation. In general, filters that are 'lossy', i.e. - the picture loses information when the filter is applied, may still be reversed but will lose fidelity. In contrast, some filters apply a hard one way function which is hard to reverse

Re:Pictures

[badasscat]

I don't find this shocking that someone was able to reverse the blure algorthum of a popular software. It is doubtfull that blur (or any other photoshop filter) was designed to be a one way function. If you know what was used, do it backwards and get the origional.

Blurring destroys image data. There is no way to "do it backwards". Let's say you have a printed photo and then you set it on fire and burn it until it's ashes. Can you "do it backwards" and somehow get your picture back? It's no different.

It should shock you if somebody was able to reverse the blur algorithm, because it's impossible. (No, "sharpen" is not the opposite of "blur". Sharpening doesn't add detail back that was lost to blurring.)

What happened here had nothing to do with reversing blur. It's just a reversed spiral. And the only thing shocking about that is that Interpol is trying to keep secret how they did it. (Oh, I dunno, select the spiraled area and apply the same filter in reverse?)

Re:Pictures

[jackpot777]

"...only under one percent of the male population is 'gay'"...?

Not even the Religious Right leaders in the US that profess an anti-homosexual [wide] stance can claim to have a gay male population under 1% in their ranks.

Seriously: don't they get Google in your compound? Just throw that little factoid out there. Right after the 50% one. Care to back it up with a URL, anyone? I know there are enough right-wingers on this forum. Easy as copy & paste, this is a nerd site, should be no problem at all. It's computers you'll be using, after all.

Unless we all want to take it as understood that the whole basis of hate-based politics is a pile of lies, of course.

Re:Pictures

[slimak]

This is pretty standard stuff in almost any junior/senior level signal or image processing course. Whether the operation can be undone depends heavily on the type of blurring. For example, if the blurring operation is linear then it can be represented by a matrix operation. This means we can write the blurred image (y) in terms of the blurring kernel matrix (H) and the original image (x) as: y=Hx (assume x and y are column vectors of all image samples). This operation can be inverted perfectly as long as H is a non-singular matrix as: x_hat = inv(H)*y = inv(H)*H*x = x.

When H is singular or non-rectangular the typical approach would be to make x_hat = inv(H^T*H)*H^T*y, which is the least-squares approximation of x from the blurred observation y (^T denotes matrix transpose).

When the operation is non-linear it generally get MUCH harder. Here, the most common approach would be to use some statistically optimal method or incorporate a priori image information.

Re:Pictures

[Sinryc]

Technically, hes right. If you like anyone of the same gender you are a homosexual. The only thing that should have made someone say "What the fuck" is that he used the word "fag". I'm just sayin'.

Re:Pictures

[IgnoramusMaximus]

Pity you did not go on, for one obvious method of showing the expertise of those "experts" would be to perform their very procedure with a picture of, say, a chicken instead of your buddy's ID, picking of course some morph stage nearer the video footage stills ... and presto! The Evil Chicken did it!

Sigh. The stuff of legal legends.

Re:Pictures

[NMerriam]

But we find it's FIFTY PERCENT. Gee... I wonder how anybody explains that...

Because pedophiles attack targets of opportunity, it has nothing to do with sexual orientation. Since there's a (near) 50/50 split of male and female children, a pedophile is going to have the opportunity to assault both sexes more or less equally.

Re:Pictures

[PW2]

a mosaic'd still photo will be impossible to enhance, but I'm expecting that someone better at math than I am will find it to be not too hard to get a very good profile of a person when working trying to de-mosaic a video -- with longer time helping with clarity

Re:Pictures

[plover]

"Magical enhancement" is possible of a blurry video image, given sufficient frames.

The trick is to use motion and multiple frames to "discover" edges. Usually the target of enhancement is a human who moves slightly, even when standing still. What the analysis software does is measure the change from each pixel to the next.

Zoom in on the pixels near a "sharply focused" curved edge in a photo of a black circle on a white background (or even a straight line on a diagonal.) You'll see filler pixels near the boundaries that are neither black nor white, but shades of gray where the individual pixel straddles the line on the original. You'll also likely see some colored pixels due to the camera sensor not having all the red, green and blue pixels at the same point. By moving the subject a mere fraction of a pixel and examining another picture, the shades and colors will change accordingly. Interpolating the difference can give you a more precise definition of where the edge really is. The more frames you compare (and with suitable motion) the more likely it is you can determine the edges.

Astronomy software exploits this fairly easily with a technique known as image stacking, as in DeepSkyStacker. However they have it fairly easy as the stars themselves make excellent reference points. It's not as easy with a subject that's not exactly trying to pose for a camera, but it's still possible.

But none of those techniques can do the whole "zoom in on that crowd, zoom in on that guy, zoom in on his name badge, zoom in on the barcode, translate that to text, look him up in the database" that spy shows like to do. There's a difference between magic and miracles.

Re:Pictures

[Beardo the Bearded]

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:Pictures

[fractoid]

Actually I was employing the 'I Did Something That Looked Similar In A Java Applet, And It Worked Like This' approach to Discussing Things On Slashdot. Mine applied a blur (actually a square convolution kernel with side length max(image_width - distance(cur_pixel, center_pixel), 0), followed by a spiral effect that rotated by t * max(image_width/2 - distance(cur_pixel, center_pixel), 0) / image_width radians, if you want jargon). It was meant for links, to look like the swirly thing that the books in Myst do when you click on them. It worked quite well. :)

Of course that was back in the heady days of "the more things that move or sparkle, the better" web design. :P I'll dig up the applet in question when I get home.

Re:Pictures

[fractoid]

Let's say you have a printed photo and then you set it on fire and burn it until it's ashes. Can you "do it backwards" and somehow get your picture back? It's no different. Yes it is. Burning the photo is actually destroying the image, because it's making it all one colour (black). Blurring an image in on computer, the way it is usually implemented, is actually a reversible operation. You can't get *all* the information back, because some information is lost due to limited colour resolution, but much of the information is still in the picture, it's just spread out. That's why on TV, when they obscure someone's face, they mash it to a 8x8 image, instead of blurring like they used to. Screen grab the blurred image and you can recover a pretty good image.

It does seem, in this case, that the image was only a twist effect, and that there wasn't any separate blur added, though. Seems pretty retarded to post such an easy-to-unmunge image of yourself fsking a kid though.

Re:Pictures

[bh_doc]

A blur does not destroy the information, it merely redistributes it. The only reason your example of the sheet of gray colour is true is because the blur has distributed the information outside the range of the image. If, however, the blur is not so drastic, then with the correct parameters a reasonably accurate approximation of the original can be produced. The general process is called deconvolution. The trick is getting the parameters right.

Re:Pictures

[hesaigo999ca]

Actually it takes less then that, they are the FBI and Adobe will always play by their rules...
they have the reverse algorythm already to undo the changes per say, so you just have to figure out the order in which the picture was changed, twist, distort etc.... no magic here, just cheating so to speak.

Re:Pictures

[zgornz]

"A blur does not destroy the information, it merely redistributes it. "

That's not true. A blur is a low pass filter, meaning it destroys the high frequency data. A very intense blur has a lower frequency cutoff than a very weak blur.

Re:Pictures

[LordVader717]

Mosaiq'd images are evry similar to blurred images - they're basically just portions of blurred, with square borders.
The key thing about reconstructing numbers is that you have a fairly limited pool of possible images, say an ASCII font.
And with both processes, one must remember that it is probably possible to reconstruct a lot more than might seem possible at first glance. The twirl effect is the other extreme, where a large amount of the original data was preserved. If he had changed the parameters a bit more to do say a hundred spins, it would be unlikely that they would have been able to make these reconstructed images, but then again they'd just look like a blur.

The best and easiest way to do it is to just black out the pieces.

Re:Pictures

[Bloke down the pub]

The only ones I've encountered were both female and dressed like penguins.

You're hereby found guilty of being islamophobic and a misogynist. Please hand yourself into your local police station for cultural sensitivity reprogramming.

Re:Pictures

[Bloke down the pub]

A blur does not destroy the information, it merely redistributes it.

Tripe. Taking an a average is a kind of blurring, and it surely destroys information about the individual values.

Re:Pictures

[ehrichweiss]

Yeah, we tried to get in touch with said forensics team and once they got wind that we were onto their methods no one ever heard from them again. With forensics like that, I can't figure out why they wouldn't want their reputation brought into question by someone who was only a hobby videographer at the time.

Re:Pictures

[bh_doc]

Counter-tripe. Taking an average is not a kind of blurring at all. A blur mixes the values of adjacent/nearby pixels across those pixels. An average produces *one* pixel from a *block* of adjacent pixels. With an average you are left with fewer values in total than the original. With a blur you are left with the same number of values.

Then, deblurring the image basically boils down to finding the inverse of a matrix that represents the blur operation. I'm not saying that you can find the inverse in every case. I'm saying that in some cases you can, and if you have the right parameters you can get pretty close to the original. I'm not saying that it's practical either, often it's not, but it's possible.

Re:Pictures

[bh_doc]

Having thought and read about it a little more, I agree with you. A gaussian blur (to define what kind of blur we're talking about) is indeed a kind of low pass filter. Although...

I'm beginning to venture into the realm of impracticality here, but in a gaussian distribution there is no zero, and so technically there is no actual information loss in a true, continuous gaussian convolution (blur). It's just that the higher frequencies are scaled down by a factor that approaches zero as the frequency approaches infinity. The trouble with this is that we don't live in a world with continuous images with infinite precision pixels. This is where the problems involved in reconstruction arise. It's not fundamental to the general blur theory, though. Which I guess is really the point I was trying to make.

I still contend that in cases with the right parameters a good deal of the original information can be retrieved, and much of the blur can be undone. I've seen it performed.

Re:Pictures

[clarkcox3]

It cannot be enhanced, but if you have some idea as to what is behind the blur, and you treat it like a hash, you can recover a remarkable amount of information:
http://dheera.net/projects/blur.php

Re:Pictures

[blahlemon]

This just goes to show you. If you are going to do something stupid, being a criminal, getting naked, etc, don't stand in front of a camera. Especially don't arrange to take the picture yourself. If you break these rule never, ever, post them on the internet or allow other people to get their hands on them to post them on the internet. If this guy is smart (obviously not) he will turn himself in. Once someone recognizes him on the street (or at work) he might be in for an ugly and sudden surprise.

Re:Pictures

[mikael]

That was the mathematical explanation of how the twirl algorithm works. If you are an image processing freak, you would write your own algorithm, but if you had a copy of photoshop, then you would use that.

Amazing technology

[alx5000]

Amazing, indeed

Re:Amazing technology

[DaveCar]

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

Re:Amazing technology

[SonicSpike]

That's the next President of the United States!

Links to the photos

[Anonymous Coward]

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

Re:Links to the photos

[aeschenkarnos]

Wow, he looks so much like a black text message saying "No Linking From This Host". I wonder if they're related.

Re:Links to the photos

[WiFiBro]

They're blocking slashdot.org originated links
Reload the link or copy it to the browers' location bar.

Interpol not the ones to descramble

[packeteer]

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

[Hognoxious]

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

Re:Interpol not the ones to descramble

[ubrgeek]

Chiefbutz, you are a madman.When you stole that cow, and your friends tried to make it with the cow. I want to party with you, cowboy.

(Yeah, yeah, OT *grin*)

Re:Interpol not the ones to descramble

[kyofunikushimi]

"unzip; strip; touch; finger; mount; fsck; more; yes; unmount; sleep" Your sig left out the gimp.

Re:Interpol not the ones to descramble

[chiefbutz]

Don't live in Germany, what would make you think that?

Re:Interpol not the ones to descramble

[torkus]

In light of the current 'one cent' stupidity on /.'s front page I think someone should charge the german police with possession of child pron. They clearly downloaded it off the interweb and AOL has the logs showing their username to "prove" it, right?

Re:Interpol not the ones to descramble

[rbarreira]

Oh my god, someone give this guy a humor injection.

Re:Interpol not the ones to descramble

[menkhaura]

Besides, there's no "unmount" command in Linux, unless you aliased it. I think you mean "umount".

Re:Interpol not the ones to descramble

[Impy the Impiuos Imp]

Yeah! "Tar" and feather him! Slay! Kill! Exec him! Stick a fork in it, he's done!

Re:Interpol not the ones to descramble

[Lars T.]

In light of the current 'one cent' stupidity on /.'s front page I think someone should charge the german police with possession of child pron. They clearly downloaded it off the interweb and AOL has the logs showing their username to "prove" it, right?

So who will claim copyright on the images?

Re:Interpol not the ones to descramble

[gold23]

Hmm. Slashdot without funny commenters is like a foot without a big toe. I think that we owe a big round of applause to our newest, bestest buddy, and big toe... ubrgeek.

Re:Interpol not the ones to descramble

[Hognoxious]

Are you claiming that you were employing sarcasm and rbarreira failed to detect it? Or that you failed to detect that I was being sarcastic, which everyone (apart from you) seems to have got?

The actual AP article

[avij]

.. can be read here.

Re:The actual AP article

[ConceptJunkie]

Wow, I wondered when /. would get to the point of not even bothering to link the articles. After all, no one reads them anyway...

Super-sekr1t unblurring techniques

[acb]

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

[Anonymous Coward]

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

[somersault]

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

[drspliff]

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

[Lord Crc]

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

I'm fairly certain they used deconvolution.

Re:Super-sekr1t unblurring techniques

[Anonymous Coward]

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!

Re:Super-sekr1t unblurring techniques

[Anonymous Coward]

In an unbounded picture with infinite intensity resolution, the point spread function of a good old fashioned blur does not destroy information. Then you can recover the image through a process called deconvolution. In the real world, information is lost at the image boundaries and due to requantization, plus you don't know the PSF. If the blur does not cover the whole image and there's only a small number of likely PSFs (as would be the case if the blur was produced in one of the popular image editing programs), two of these deconvolution limitations are already gone and you're only dealing with quantization noise, which isn't so bad if you're only looking for high contrast information like number plate letters. It gets easier if you have multiple pictures or even video.

Re:Super-sekr1t unblurring techniques

[Dan100]

Interpol's top-secret image-unblurring technology

Nowhere do Interpol suggest that they were using "secret techniques".

Re:Super-sekr1t unblurring techniques

[petermgreen]

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.
unforuntately that probablly won't work because there is almost certainly interpolation going on.

so you need to run lots of test images to try and calculate how much influence each pixel in the untwirled image has on each pixel in the twirled image.

but if you have lots of computer power at your disposal and the tool in question is scriptable (or you can bully a scriptable version out of the app vendor) then this shouldn't be too hard.

Re:Super-sekr1t unblurring techniques

[WillAdams]

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

[Stradivarius]

Actually you can do the same sort of operation to unblur a license plate. In fact that was a lab exercise in one of my college signal processing courses :-) The main concern I'd see with the license plate problem would be ensuring your camera sampled the data fast enough to meet the Nyquist criterion. But if you did, you'd have all the information content in the image you'd need.

The degree of blurring isn't really the issue, unless you start to lose information due to quantization losses. Short of that, it's a matter of A) being able to guess or approximate the original blur operation, and B) the fact that "blur" operations are generally reversible (they're simple linear shift-invariant filters, and you're not losing information in that process). So you just apply the inverse filter to the image (or subset of the image). If you don't know the exact original blur, you can attack the problem by brute force and just try inverses of many different blurs until one looks fairly good.

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). That's far more reliable than applying even the heaviest of blurring filters.

Re:Super-sekr1t unblurring techniques

[gedeco]

Correct me if I'm wrong, but would mirror the picture and applying the twirl effect in the same direction again not be exactly the same?
All you need is the same program that was used to create the twirl effect.

Re:Super-sekr1t unblurring techniques

[sqldr]

Next they'll be doing them as slide puzzles.

Re:Super-sekr1t unblurring techniques

[richlv]

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

[1u3hr]

Nowhere do Interpol suggest that they were using "secret techniques".

It wasn't Interpol who did the descrambling, it was the German police. However, they did imply they wanted to keep the methods secret:

Germany's Bundeskriminalamt, or BKA, police force. Contacted separately by The AP, a BKA spokesman said the agency did not want to detail the process used by its image processing expert "because we do not want to give criminals the opportunity to adjust to the techniques we are using."

Re:Super-sekr1t unblurring techniques

[tgd]

Did you save in-between?

Photoshop may have been keeping some amount of subpixel and original image data in memory that let it do that. I *think* it does as it allows undoing of filter effects and avoids the loss of quality when layering filters.

Re:Super-sekr1t unblurring techniques

[The New Andy]

I don't believe the twirl effect is a convolution. Convolution effects are things like gaussian blurs, where each pixel in the image is modified the same way regardless of where it is in the image (e.g. each pixel takes on the average of the 8 surrounding pixels).

Deconvolution will do some unblurring of images, but obviously it can't synthesize information that has been removed.

Re:Super-sekr1t unblurring techniques

[FuzzyDaddy]

(they're simple linear shift-invariant filters, and you're not losing information in that process)

Which points to the difference between the computational and physical worlds. If I blur an image by, say, projecting it through a lens on a piece of film, I reduce the amplitude of the high spatial frequency components. As you say, this isn't a lossy process. The difference in the physical world, there's noise added at every step, and the physical blurring reduces the signal to noise in the higher spatial frequency components, causing some true loss of information.

Doing a blur in photoshop reduces the amplitude of the high spatial frequency components as well, but there is no added noise - so the process is reversible.

Re:Super-sekr1t unblurring techniques

[RatCommander]

I'm fairly certain they used deconvolution.
No, I'm fairly sure they didn't. Deconvolution is only applicable to linear systems - this "twirl" filter is non-linear. A better bet would be something like the reverse mapping approach mentioned elsewhere in this thread.

Re:Super-sekr1t unblurring techniques

[paulatz]

You are not completely accurate: some kind of filters you can apply to an image are reversible, while some other aren't. Even a blur can be reversible or irreversible depending on how exactly it is applyed. E.g. you can FFT the image then cut away the higher frequency than FFT it back, this kind of blurring is irreversible as the higher requency data is lost forever. Another way to blur (the most common one, actually) is convolve the image with a gaussian, this mean that you replace every point-like pixel with a little gaussian, then sum all the gaussians trasforming it back to pixels; this algorithm is in theory reversible, even if you loose some data due to numerical round-off and space discretization. Of course you can always draw a big black square over the zone you wish to hide; this is definitely irreversable.

Re:Super-sekr1t unblurring techniques

[jamesshuang]

Why not just use the simplest one - a black box? Information can't be extrapolated from something that doesn't exist. Also, I doubt you'd have problems with recognition - if anyone sees a body, neck, and a black box, I doubt they'll think you have decapitated people running around...

Re:Super-sekr1t unblurring techniques

[meeya]

I think you are right.that's
what I thought too.unfortunately I don't have any of these software to try that.

Re:Super-sekr1t unblurring techniques

[GregNorc]

I like to paint in a tone similar to whatever I'm covering... then add some noise and a blur. I like the idea of some self fashioned internet detective spending 45 minute to uncover... nothing.

Re:Super-sekr1t unblurring techniques

[YrWrstNtmr]

Now, taking this several steps farther, Interpol had to do it on pics grabbed from the web, of unknown compression, originally done with an unknown application, possible 3rd party twirl filter, and saved/compressed unknown number of times at unknown quality level.

Still, the (alleged) perp is still a huge dumbass.

Re:Super-sekr1t unblurring techniques

[4D6963]

A gaussian blur is indeed a good choice, as it's pretty efficient at completely "destroying" the higher frequency components (finer details) you want to definitely lose, although filtering with a windowed sinc (aka Lanczos filter) is even more efficient at doing that.

That is to say, straight "mozaicing" with no prior filtering lets you opportunities to reconstruct the original image using superresolution techniques (only in the case you show more than once picture of the mozaiced subject, such as in a video)

Re:Super-sekr1t unblurring techniques

[SatanicPuppy]

For something that would pixelize, I'd think it would be less reversible, because there would be a certain amount of "rock and roll" (random data) added to the effect...But for something like "swirl"...I'd imagine that effect is mostly a mathematical transform.

The only place you'd lose data is in the rounding process.

Re:Super-sekr1t unblurring techniques

[Mr Z]

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

[Mr Z]

...errr... Naturally, I don't paste faces for the numbers... ;-) YouknowwhatImeant.

Re:Super-sekr1t unblurring techniques

[AmiMoJo]

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

[andr0meda]

See, this is EXACTLY is why you eventually skipped grades.

Re:Super-sekr1t unblurring techniques

[corsec67]

Which is why I paint over areas to be hidden in a solid color that I eyedropped from that area. Then there is absolutely nothing you can do to recover data, and all you know is the color of one pixel in that area.

Re:Super-sekr1t unblurring techniques

[Hoi Polloi]

Why did this guy blur his face instead of just erasing it? He didn't want to ruin the mood of his kiddie porn?

Maybe if he had put "copyright" on it he could sue interpol.

Re:Super-sekr1t unblurring techniques

[Random832]

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

[CmdrGravy]

Actually not, the process used is much more advanced and intensive than that.

First of all a set is built in a warehouse to mimic the location shown in the photograph, prisons are then scoured for inmates possessing a similar height and build to the suspect and placed on the set in the correct position. A special "twirling" device is then employed to physically "twirl" their head to correct angle as that used in the original photograph and a photo is taken. This photo is then checked for a match with the original, if there is a match then our suspect will have been caught. Unfortunately he is now dead.

Re:Super-sekr1t unblurring techniques

[pclminion]

In theory all convolutions are completely reversible, including FIR blur, even very extreme blur. In practice, though, the pixels of an image have finite depth of color and information is lost during convolution. If there was such a thing as "Infinite-bit color" (as opposed to say, 24 bit color), then it could be reversed perfectly, no matter how blurred it was.

For IIR blur, where the kernel is basically infinitely large, it is no longer possible to exactly reverse the blur, because the result is infinitely large (thus the first "I" in "IIR"). Since all real images are finite, this means a lot of necessary information is cropped away and the blur cannot be reversed.

Re:Super-sekr1t unblurring techniques

[pclminion]

Why do you bother? Why not just cover the face with a black rectangle and remove all possibility of ever uncovering it?

Re:Super-sekr1t unblurring techniques

[somersault]

If a random component were added to the blurring process it wouldn't be reversible, though I guess it would then be more of a 'scramble' than a blur..

Re:Super-sekr1t unblurring techniques

[pclminion]

I'm fairly certain they did not, given that the twirl effect is not a convolution effect.

Re:Super-sekr1t unblurring techniques

[MenTaLguY]

This is true as far as it goes; but while perfect deconvolution is not generally possible, you can often get it "good enough".

Re:Super-sekr1t unblurring techniques

[ArsonSmith]

It makes the image 2d rather than 3d.

Re:Super-sekr1t unblurring techniques

[mattOzan]

Show us pictures of your wife!

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

Re:Super-sekr1t unblurring techniques

[butterwise]

they did imply they wanted to keep the methods secret

Why bother? I sure won't be using that method to mask my identity ever again...

Re:Super-sekr1t unblurring techniques

[OrangeTide]

Or just don't take photos and post them, then there is even less information to extrapolate. Criminal's ego and arrogance tends to get their ass caught.

Re:Super-sekr1t unblurring techniques

[diskis]

Nope, they just untwirled it.
Open the image in photoshop, center a selection in the middle of the face, filter -> distort -> twirl, amount to -999, and there ya go.

Re:Super-sekr1t unblurring techniques

[hughperkins]

There used to be a couple of before-after pictures in the physics laboratory at Cambridge of removing blur from a slow-shutter picture of a car going past.

- in the before, the licence plate is blurred out and unreadable
- in the after its perfectly readable

They were using fourier analysis to do this. My tutor was doing a phd on this.

Re:Super-sekr1t unblurring techniques

[K-Mile]

Wow, it really works! I took a picture of my girlfriend, did a -999 degrees twirl, and she now looks perfectly normal! Amazing technology...

Re:Super-sekr1t unblurring techniques

[Lord Crc]

No, I'm fairly sure they didn't. Deconvolution is only applicable to linear systems - this "twirl" filter is non-linear.

I admit I'm no expert, but from what I can see the twirl effect can be modeled as a spatially variant PSF, and from what I can see, deconvolution can handle this.

Re:Super-sekr1t unblurring techniques

[maybenot]

Nope, just opened it up in photoshop and tried twirl and fooled around with reversing it, not that simple. They did something else.

Re:Super-sekr1t unblurring techniques

[IgnoramusMaximus]

Err ... the GP poster was working with images of patients for a medical journal. Unless of course you are insinuating that getting ones articles printed in an overpriced medical journal in itself constitutes a crime ...

Re:Super-sekr1t unblurring techniques

[cyberstealth1024]

There are ways to determine (or approximate) some of these unknowns...

• Quality level (approx): http://newsgroup.xnview.com/viewtopic.php?t=10298#37861 ..of course you can't determine how many times an image has been saved at whichever quality level(s).
• Application: possibly in EXIF if the application put it in automatically, and if the user didn't know to scrub it
• Compression level, 3rd-party twirl filter: dunno workarounds for these

It definitely makes it harder with the unknowns...put obviously not impossible. I concur with the others by saying "data masking != data removal"

Re:Super-sekr1t unblurring techniques

[JuliaNZ]

And if you do that then it's even easier to recover the original, just open it up in GIMP and remove the layer you drew the black boxes on. "Flattening" an image means getting rid of all the separate layers and making it one single composite image, which is what you'd want if you were trying to hide something.

Re:Super-sekr1t unblurring techniques

[OrangeTide]

I'm just saying that he has an ego and is a criminal. I didn't mean to apply I know what crimes he committed. "Patients" could be victims for all we know.

Re:Super-sekr1t unblurring techniques

[munpfazy]

Somehow, I have a feeling the parent poster is aware of that.

Blurring different from twirling...

[Max Romantschuk]

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...

[ultrasound]

Surely the most effective filter would be a GIMP mask?

Re:Blurring different from twirling...

[Joce640k]

Next time remember to use the "mosaic" filter....

Re:Blurring different from twirling...

[KlaymenDK]

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...

[zebslash]

Exactly, contrary to a blur or a crude mosaic, a "twirl" is almost non-destuctive, and therefore it is possible to put the pixels back in place.

Anyway, many people will think it is an amazing achievement and that it is possible like in many movies, to get a very fine, detailed image from an out-of-focus, low-res picture.

Re:Blurring different from twirling...

[baldass_newbie]

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

Even better would be to not rape little kids in the first place.

Re:Blurring different from twirling...

[hcdejong]

Or do it in hardware instead: Black bar glasses.

Re:Blurring different from twirling...

[speaker of the truth]

Anyway, many people will think it is an amazing achievement and that it is possible like in many movies, to get a very fine, detailed image from an out-of-focus, low-res picture. I saw it done in CSI so it must be true, because CSI isn't a movie!

Re:Blurring different from twirling...

[DrXym]

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

I doubt it is. I recall an article some time ago that showed you could guess words that had been pixellated by statistically analysing what the result looked like. While the same may not be strictly true of a human face, I would not be surprised if they could work out your general features, hair colour, eye colour, and facial features. Enough to produce an artists impression.

It seems if you really want to hide what you look like you need to blank out your head entirely as well as any possibly incriminating marks on the rest of your body.

Anyway, this asshole deserves to be caught and I hope they do catch him.

Re:Blurring different from twirling...

[Sir Holo]

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

Perhaps for a photo, but not in a video clip. It's likely you can find 2 seconds of video in which the face does not change orientation, particularly if they are moving around. Moving around changes the binning of the image into a pixellated one. You can reverse it by frame-averaging 2 seconds (30 frames) of a blurred face, which in effect is oversampling a source, and you can get enough to recognize a face.

Re:Blurring different from twirling...

[taylorius]

"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"

Not true actually - at least theoretically, a gaussian blur (or any convolution) amounts to nothing more that a change of basis - and that change can be undone.

However, the need to know the EXACT kernel used, and the quantized (8bits per channel) data, in the case of an image, may well render it practically impossible.

Re:Blurring different from twirling...

[eimikion]

Actually parent is right. There were no rape report from anyone. Who are the alleged victims? Do they testify against someone?

Re:Blurring different from twirling...

[apparently]

You don't need a rape "report", if there are images of the assailant molesting a fucking kid, numbnuts. Are you attempting to play some half-assed version of Devil's Advocate, or are you just that ignorant?

Re:Blurring different from twirling...

[UbuntuDupe]

That's one of the few times a /. post has made me literally laugh out loud. Thank you sir :-)

Re:Blurring different from twirling...

[apparently]

Somewhere in there, you really should have articulated a point specific to my post, instead of some odd rant against "Perverted Justice" that has nothing to do with what I said. Let's look at the specific issue again:
There are pictures of a grown man molesting children.
Your argument is that pictures of a grown man molesting children isn't sufficient evidence for that grown man to be prosecuted; you require the molested children to file a report.
You're a fucking ~brain~. Do you have a newsletter I could subscribe to that could enlighten me to the expanded worldview where pictures of child molestation occupies a place in some moral gray area?

Re:Blurring different from twirling...

[pclminion]

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.

This statement is false. A convolution is a linear option which maps N degrees of freedom to N degrees of freedom, and is perfectly reversible (in theory). In practice, the quantization of pixel values makes it somewhat lossy, but still reversible, although not perfectly.

Even if you used an infinite blur (IIR gaussian or something), most of the relevant information remains in the image. Blur is easily reversed.

As a physical analogy, imagine projecting an image through a lens which blurs the image at a certain focal distance. This blur can easily be corrected simply by using a second lens. Information is not lost in the blurring process, it is just spread around somewhat.

Re:Blurring different from twirling...

[apparently]

I have said nothing regarding statutory rape of the classically defined "19 yr old with a 17 yr old" scenario. I have spoken to the specific issue of adults molesting children. And even more specifically, to the issue of their being photographic evidence of said adult molesting said child, in which the original post implied to as not being sufficient evidence.
"The intent here is to intimidate and discourage discussion"
Please.
You're discouraging discussion by broadening the topic to include sex between a 19 and a 17 year old, while the discussion at hand is specific to child molestation.
You're discouraging discussion by inferring that because I'm against adults molesting small children, I don't think a moral code can exist that punishes child molestation while not punishing teenage sex.
You're discouraging discussion, because you're using my specific point, to argue against a POV that I do not hold.
A discussion cannot be had, if you insist on jumping off on a tangent and start attacking an argument that I'm not even making in the first place.

Re:Blurring different from twirling...

[apparently]

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.

Re:Blurring different from twirling...

[LadyLucky]

> 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. Incorrect. A pure gaussian blur is 100% reversible, except when it is applied near the edges of images (since you would need information off the side of the image). The presence of noise can make the gaussian blur difficult to undo, but nonetheless all the information is retained in the blurred image - there is only one image that will produce exactly the blurred function you see.

Re:Blurring different from twirling...

[R2.0]

Sorry, you are on the wrong end of this argument on a couple of accounts:

1) Ad hominem attack, invoking the OP's "set aside their emotions and have a rational discussion on this topic", as well as the first response (possibly not you, but definitely sympathetic) invoking a religious bias that was not in evidence - no invocation to God was made, and "evil's advocate" has been a thoroughly secular phrase for hundreds of years.

2) You are attempting to invoke the spectre of censorship of discussion when teh discussion has no bearing on the topic. "the fascination discussion around the actual origins of consent and rape laws" really has no bearing on the topic at hand. There are pictures of an obviously adult make (pushing middle age) engaging in sexual acts with individuals who are obviously pre-pubescent. Such young individuals CANNOT, under any modern (or even archaic, to my knowledge) legal theory, give consent to a sexual act. Those photos are prima facie evidence that a rape has occurred, unless we want to wander off into NAMBLA land. Now, if they find this guy, and he can show that the pictures were faked, he had a gun to his head, they were actually 40 year old midgets, then he could argue his case. But to a reasonable person, those pictures show a rape occurring.

3) You are picking the wrong example to argue your case. If indeed the subject had been having sex with pubescent or older youths, one might argue that consent was given, etc. There are indeed gray area in law and morality. This is not one of them. You are proposing a classic slippery slope arguement: because some young people might be able to consent to sexual activity, ALL young people might be able to consent http://www.don-lindsay-archive.org/skeptic/arguments.html#slippery I could just as easily take the convers: because the very young cannot give consent, than no one under the age of 21 can give consent. You reject that argument, butit is just as specious as the one you are making.

In short, you've picked a loser. You intend your arguments to make you appear reasoned and rational; instead you come off as hair splitting and equivocating.

Come back with a better subject next time. Good day to you.

Re:Blurring different from twirling...

[R2.0]

Sorry - typo on the "evil's advocate" - it was supposed to be "Devil's advocate". Fat fingered the CTRL key.

Re:Blurring different from twirling...

[apparently]

Here's a concept for you: Read The Fucking Article.
Since you need some help with click-clicking your way around the interweb, here's the relevant sentence:
"INTERPOL is seeking the help of the public to try to identify this man, photographed sexually abusing children in a series of images posted on the Internet."

You ask: Where are these supposed pictures of an adult "molesting" a child? All I can see is 4 pictures of an adult sitting around...

Correct. Interpol didn't post the explicit pictures, probably because posting pictures of the molested kids isn't necessary. Instead, they posted images that they needed the public's help with, that is, headshots of the assailant.

And you state: Oh, and "alleged" because that's his status until he has been convicted of a crime by a judge

I'm not referring to the legalese definition of "alleged". Are you telling me that when given incontrovertible proof of a crime taking place, your thought process is "Gee George, that only looks like an alleged crime. Can I go pet the pretty lady's hair now?", you fucking Lennie?

Hardly Rocket Science

[clickclickdrone]

>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

[zebslash]

It does strike me as a bit stupid explaining it all - now crims will just use better techniques for blurring themselves out. I think that most technical people (as found here on Slashdot) will have figured out how they did it. Many posters here gave the solution (as the "twirl" is not a very destructive transformmation). If Police had not explained how they did it (and actually they tried to add up some mystery to their story by the way), most journalists would have commented on the method anyway.

Re:Hardly Rocket Science

[suv4x4]

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. 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

[shilly]

The police said that they the cost of revealing they can do this was worth paying given the potential prize of tracking this guy down. They made a trade-off.

Re:Hardly Rocket Science

[locster]

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.

Re:Hardly Rocket Science

[suv4x4]

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.

Re:Hardly Rocket Science

[houghi]

So you rather have security through obscurity?

Re:Hardly Rocket Science

[trmatthe]

I think you are wrong. When I saw this article I immediately thought "Just reverse the process". I am pretty sure I'm not a genius, so this occurred to many people (as evidence by the number of posters here). We aren't going to improve security by trying to hide the vulnerabilities. It's called Security through Obscurity and has been widely discredited.

What we should do is nail this guy which may act as a deterrent to some potential paedophiles. With all the tools in their arsenals, I don't think Interpol are giving away the crown jewels in intelligence here.

tim

Re:Hardly Rocket Science

[clickclickdrone]

>With all the tools in their arsenals
Somehow this seems the wrong turn of phrase given the subject matter.

Re:Hardly Rocket Science

[jmv]

While it *might* be possible to recover a Gaussian blurred image, it wouldn't be possible to do anything if the image was blurred with a sinc function. Even adding a tiny bit of noise after the Gaussian blur would most likely prevent reconstruction (because any attempt at deconvolution would amplify the noise).

Re:Hardly Rocket Science

[MikeBabcock]

One of the smarter ways to apply pixelation is to limit it to a fixed colour map, especially in situations with multiple photos or in video, since multiple frames allow greater data recreation otherwise. With a small fixed palette, the unpixelated image becomes even more impossible to reconstruct because of the permanent loss of accurate colour data.

Re:Hardly Rocket Science

[poot_rootbeer]

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.

Yes, but that extra 10-20% of quality could mean the difference between making a positive ID and letting a monster go free. Worth the extra effort, no?

now crims will just use better techniques for blurring themselves out.

Criminals who let their faces be photographed in the act are pretty stupid to begin with.

Re:Hardly Rocket Science

[GuyinVA]

I use super secret MS Paint to alter my images. I 'blur' it by using the paintgun. Then resave it as a new file so the original data isn't kept as part of the new file... Of course, my blurs are just for stuff like license plates, account numbers, etc. I hope future criminals aren't as simple as me.

Re:Hardly Rocket Science

[Jah-Wren Ryel]

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", It's not insane. They know that movie-plot terrorist plans are a dime a dozen and since we've had precisely one terrorist attack in the USA for in over a decade, the chances of anyone actually using one of these plans is just about nil.

But this way they get to generate a lot of publicity, justify their jobs, gynormous budgets and/or advertising dollars and keep all the sheeple reminded of just why they need the government.

Re:Hardly Rocket Science

[Nocturnal Deviant]

Well at least they spun it for good reasons...the other guy was spinning meatspin style

Well I personally welcome our new "not using Photoshop, using a special technique we wont release" German interpol overlords. They'll do a better job than our old robot overlords...er actually maybe not interpol has politicians in it http://topics.nytimes.com/top/reference/timestopics/organizations/i/interpol/index.html?query=POLITICS%20AND%20GOVERNMENT&field=des&match=exact /cry, DON'T LEAVE US ROBOTS WE 3 YOU

Re:Hardly Rocket Science

[32771]

Not that it would be entirely on topic but you just remembered me of this guy who took high resolution pictures of the ISS, the Space shuttle, some spy satellite, and Mercury. He took multiple frames of the object of interest and selected the best for combining well here it is what he exactly did:

"Images of Mercury were obtained at 8 bits of resolution using exposure times of 16.7 ms at a rate of 60 frames per second and recorded on broadcast-quality videotape for subsequent data reduction. The images were sorted and selected based on maximum gradient of the planet's bright limb, co-aligned, and added in 16 bit space."

Here is a link:

http://www.journals.uchicago.edu/AJ/journal/issues/v119n5/990240/990240.html

let me store it here so I don't forget it ;)

This method is used to reduce seeing which is random and capable of reducing that Mt. Wilson telescope aperture from 1.5m to some seeing limited aperture of maybe 10 to 20cm (my guess).

The pixelation problem is not that similar to seeing. I would think since each new larger pixel is the average of the same region in the original image it is somehow low pass filtered spatially. Recovering a part of the image would somehow make it necessary that the image information is stored in the time domain since you can't get it from that 2d single frame space. That high level view makes it again look like the seeing problem but with the seeing one gets some images which cover a large spatial bandwidth but have low dynamic range, while with the pixelation the spatial bandwidth is constantly low but the dynamic range in the image is constantly high.

Searching google gives me something about de-identification and how the simple methods here discussed are easily thwarted by face recognition software. I.e. the bad guy crosses the US border is photographed, later produces de-identified compromising images of himself. Then the blurred/pixelated image is fed into the face recognition program and compared to the border database - success should easily follow, because the facial features are still recognizable to the software.

Here is an example: http://reports-archive.adm.cs.cmu.edu/anon/2003/CMU-CS-03-119.pdf

I just can't find what you were talking about, and I'll come across who knows what if I try to find it myself:

http://www.cl.cam.ac.uk/~mgk25/ieee02-optical.pdf

Re:Hardly Rocket Science

[32771]

Ha! Found it!

http://citeseer.ist.psu.edu/393340.html

There is a paper I can go to bed with ;).

Re:Hardly Rocket Science

[locster]

Indeed :)

Looking through the paper about reconstructing a CRT image from its reflection from a diffuse surface (or wall to normal people) - I was pretty amazed at the final result. And that name in the image, Markus Kuhn, hmm, ahh yes he was the guy who cracked videocrypt - the system used by BSkyB (and others) to scramble their analogue satellite TV video output back in the late 80's and 90's. I know 'cause I watched a good many hours of TV for free using versions of his code ported to the ole Commodore Amiga. Small world eh :)

a better solution

[User 956]

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.

Re:a better solution

[15Bit]

Sadly they will CONTINUE molesting children and STOP photographing it. It pushes against my freedom of information beliefs, but in cases like this i can see the argument for not telling the world how you solved the crime...

Re:a better solution

[SamP2]

People who are not idiots (well assuming there are any non-idiots that molest children in the first place) won't be dumb enough to record it on camera to begin with.

The idiots will keep doing it regardless.

Status quo.

Re:a better solution

[suv4x4]

--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.

And before you know, you've become part of the "think of the children / if you've done nothing wrong, you've got nothing to hide" crowd.

But of course, if we try to protect our identity, we must molest children.

Re:a better solution

[IndustrialComplex]

Sadly they will CONTINUE molesting children and STOP photographing it. It pushes against my freedom of information beliefs...

Oh dear. I don't think I need to mention how that could be unintentionally misread.

Re:a better solution

[yoprst]

well, actually not using fancy photoshop filters instead of good old black squares is safe enough

Re:a better solution

[speaker of the truth]

Could you post how it was meant because all I'm doing is misreading it (and I realize I'm misreading it, I just can't work out the proper reading of it).

Re:a better solution

[IndustrialComplex]

He was referring to publishing the method used to de-whirl the images, not the images themselves.

Re:a better solution

[speaker of the truth]

Aaah okay. I was thinking he meant he can understand why criminals wouldn't publish photos of themselves because it would get them caught.

blurring != obscuring; true, but...

[Animaether]

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/ ) 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.

Re:blurring != obscuring; true, but...

[Splab]

I can't help but think people using blur/twirl/distort is somewhat trying to get caught, I mean if I wanted to make sure no one was able to see my face I would cut the information out of the picture - mark it, delete - it goes black and there is no way back.

Re:blurring != obscuring; true, but...

[Aokubidaikon]

you're not going to be able to just take any heavily non-motion blurred image and get a supersharp result back.

Clearly you need to watch more episodes of CSI ^_^

Re:blurring != obscuring; true, but...

[weierstrass]

Maybe you want to be able to prove that's it's you in the photo. What the police have done has proved that the swirl isn't a one-way function - but if you thought it was you might want to produce the original photo and show someone that it was really your face.

Re:blurring != obscuring; true, but...

[ockegheim]

Yes, with focus blur, you have the same number of pixels as you would otherwise have, just mixed in a way that isn't immediately obvious. Though I suppose if the Hubble Space Telescope required corrective lenses, it mustn't be all that easy.

How to remove numbers/faces from a picture

[kasperd]

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.

Re:How to remove numbers/faces from a picture

[chuckymonkey]

Honestly though, how many people think like that? Most people are not going to be aware of an organization's ability to de-obfuscate an image and will not take the appropriate steps to hide their identity. The only thing I worry about in this post is the lack of peer review in source code, I believe that if the government is going to use software that may have a negative impact on a person's life (not saying this guy is innocent, just in general) then the source needs to be freely available for peer review so that the margin for error can be out in the open and the quality of the code verified.

Re:How to remove numbers/faces from a picture

[twoshortplanks]

So, how about instead of blurring people's faces in Canada Google replace them all with someone's face. The Google founders seem like a good idea, but John Malkovich would be even funnier.

Re:How to remove numbers/faces from a picture

[edittard]

You just have to say "zoom" and "enhance". See, all that time watching CSI wasn't wasted.

Re:How to remove numbers/faces from a picture

[Asic Eng]

It's the judge's job to handle this, if the guy is found and brought to court. If the picture is a major piece of evidence he may order the source code (or the sequence of actions they executed in photoshop) to be examined. I suspect though that the picture would only be used as a way to track this person. As a starting point to gather more information, not as acutal evidence in court.

Re:How to remove numbers/faces from a picture

[pclminion]

I believe that if the government is going to use software that may have a negative impact on a person's life (not saying this guy is innocent, just in general) then the source needs to be freely available for peer review so that the margin for error can be out in the open and the quality of the code verified.

Are you seriously suggesting that a bug in the code could cause the software to reproduce some INNOCENT person's face instead of the real face?

Anyway, your naivete is sort of amusing. If the government says "Yeah, we're using software X, here's the source code for you to look over," are you actually going to believe they are really using that code?

Re:How to remove numbers/faces from a picture

[chuckymonkey]

Yeah, my naivete is amazing considering that I have worked in those circles. You'll also noticed I never said anything about it actually happening, just that was how I think things should be. It's also known as an opinion and it's neither right nor wrong nor even naive.

You realize why they are saying all this?

[Anonymous Coward]

2 reasons.

1. The average person out there really does not understand computers. As such, the criminals look smart, and now the police look even smarter. It is hoped that by making the cops appear to be intelligent that other criminals will stop as well as perhaps the cops will get more money for this.
2. Chances are, that the cops still did not understand it and could not explain it if they tried. More likely, they took the pics to a geek and they did the reverse. Notice the web site? I am guessing not a person there has 1 clue.

Re:You realize why they are saying all this?

[suv4x4]

The average person out there really does not understand computers. As such, the criminals look smart, and now the police look even smarter. It is hoped that by making the cops appear to be intelligent that other criminals will stop as well as perhaps the cops will get more money for this.

Fear of the unknown is a better weapon, than giving forensic analysis tutorials to the entire world.

And what they achieve is they look dumb now, since anyone having a clue knows the transform is basic. It may push some smarter people doing a crime since they believe if the police is so proud with their untwirl, they must be on a pretty low level overall.

I'm not saying this will make smart people molest children and shoot photos of themselves, but still, your reason is weak..

Re:I bet it's made up

[geomark]

Very well could be made up. Look at how police are compensated. They get rewarded for closing cases. The focus is on arresting a suspect and getting a conviction. Whether or not it's the right person is not part of their pay package.

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

[Gordonjcp]

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...

[maroberts]

You're right, but if someone knows what package was used to edit the pictures, the filter might also be known, in which case there may be a mapping process back to the original.

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

[Gordonjcp]

No, there isn't a mapping back to the original. There just isn't. It's like trying to work out what two numbers I started with if I say "4" - was it 2+2, or 1+3, or 0+4? You've no way of knowing.

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

[cnettel]

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.

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

[ardor]

You don't have to. Just extract the facial features (which are unique), and match them against a set of known photographs. Blurring does not remove them unless you blur a LOT. The only reliable way of hiding the identity remains the black bar in the actual pic (not just superimposed).

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

[bentcd]

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. If the result after blurring is in some way dependent upon the original image, then:
1) If you have a list of suspects, you could apply the same algorithm to each of them and see which one best matches the picture.
2) If you have a blurred-out video sequence, you have a lot of time data to work with. As the camera pans, or the person moves about, different pixels will get blurred in different ways and this will happen in a way that depends on the data of the original picture. Capture enough of these frames, and you /might/ be able to work yourself back to reconstructing one single picture that is reasonably close to the subject's actual appearance. You will definately not be able to reconstruct each and every one of the individual frames in the sequence but that's probably not necessary anyway.

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

[sim82]

Well I guess the prior information that there was something that looked like a face under the blurred blob could be used for some kind of statisitcal estimation of how the face might have looked. I think that much less information is needed to extract biometric features from a blurred picture, than to do real un-blurring. Has something like that been used in practice?
It could work somehow like the 'de-pixelization' of license plates where AFAIK the most likely licenseplate number is estimated instead of doing real (impossible) de-pixelization.

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

[IndustrialComplex]

Not when you phrase it like that. Of course you can't get the 'exact' data back, but you can get an approximation. And think about what a digital picture is in the first place. It is an approximation of an image in a pixelated form.

Exact doesn't matter in terms of identifying a person via a photograph. While 'how' exact is left to a jury, you cannot claim that ANY picture is an exact reproduction of the original image. There is ALWAYS loss.

The photograph or in this case, unblurred image only needs to be close enough. And 'enough' is defined by the person who will use the image. The user could be a jury, an investigator, a judge... each would have their own definition of how important the unblurred image is.

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

[petermgreen]

true, however a blur is not merely doing that, a blur is at it's heart convoloution with an inpulse response. If you know (or can guess) the blur function and where it was applied then you can come up with an inverse.

on the other hand a randomly applied blur tool or a block averaging tool is going to be much much harder.

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

[Goaway]

So deconvolutions don't exist, now?

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

[Datamonstar]

If they didn't this post wouldn't be echoing right now.

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

[26199]

It's interesting to note that given enough precision a blur is fully reversable.

In the frequency domain, a guassian blur reduces the amplitude of high frequencies; it doesn't drop them to zero. Simply multiply up again to get the original image.

Now, in practice this doesn't work particularly well because the high frequencies end up with such low amplitude that quantization destroys them.

The reason there is still a potential problem is that the 0-255 precision used on computers holds more information than the human eye can fully perceive. So it may be possible to unblur things when, to a person, the information is gone.

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

[Gordonjcp]

*However*, adjacent pixels will have similar equations.
... and it's just these similar adjacent pixels that lossy compression will compress out. So, when you get your image, the information has been well and truly lost.

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

[Urban Garlic]

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

I applied my own personal proprietary seekrit algorithm, and got "3", "1", "3", "3", "7", ":", ")".

What do I win?

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

[Goaway]

Deconvolutions still work on JPEG images, you know. You lose some information, but certainly nowhere near all of it.

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

[smellsofbikes]

>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.

Almost like how an MD5 of an *enormous* file is a very small number, and the MD5 of another enormous file is a *different* very small number, and if you have two enormous files that produce the same MD5, they're probably the same file...?

Just sayin'.

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

[Goaway]

Strangely enough, I can't fix bugs through telepathy, so how about reporting it first?

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

[Goaway]

Put it on http://code.google.com/p/xee/issues/list, or I'll forget about it.

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

[Gordonjcp]

Not really, no. Not much like that at all.

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

[smellsofbikes]

Oh really?
If I take a TIFF of Angelina Jolie and another of Laura Bush, and run them both through a lossy transformation algorithm known as gif, I can sure tell from the result which starting file yielded the final file. If I take a WAV of JS Bach and another of Metallica and mp3 them both, I can clearly identify the starting file from the result. When I was younger I spent a lot of time logging onto systems that had /etc/passwd visible, downloading it, then running dictionaries through crypt to find matches. The basis of all hashing is that reduction in data volume does not equal irrevocable loss of data->hash identity.

Are you saying that a Photoshop swirl is somehow magically different than all these cases? Would you care to support that assertion with, y'know, *anything*?

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

[Gordonjcp]

Are you saying that a Photoshop swirl is somehow magically different than all these cases? Would you care to support that assertion with, y'know, *anything*?

No, and if you actually *read* my original comment you'd see that I wasn't talking about a swirl effect at all. Reversing a swirl effect is trivially easy. In fact, we spent about an hour at work swirling and unswirling images just for a laugh, because we're easily amused (and you do get some really odd faces).

What I said is that *blurring* the image is not a generally reversible process. You can't put back high frequencies that aren't there any more. If you use a convolution blur *and* you know the impulse response then in theory you could reverse the blur and get back the original image. If you apply any kind of blur (convolved or arithmetical averaging) and lose some of the data, you're unlikely to get the image back. If you blur and then use a highly lossy format like JPEG, then it will go "See that bit? That's just one big pinkish-beige blob all the same colour", and the fine detail will *definitely* be gone.

If I take a WAV of JS Bach and another of Metallica and mp3 them both, I can clearly identify the starting file from the result.

If I take a WAV of Bach and stick it through a lowpass filter, chopping off everything below 100Hz, and then MP3 it, can you even identify it as music? Can you get the music back? Why not?

In other news...

[djupedal]

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

New laws were passed today, making it a felony to obscure, obfuscate, scramble, cover or otherwise purposely mask your identity by modifying a digital image for the purpose of avoiding identification by law enforcement agencies.

Re:In other news...

[Tim C]

Actually, I could see someone trying to make a case that that's either obstructing the police or attempting to pervert the course of justice, assuming that you were caught for whatever it was that you're accused of.

Don't forget though that a number of things that are legal (eg carrying a crowbar while out and about) become illegal if you are engaged in a related crime (carrying that crowbar while breaking into a house becomes "going equipped").

Re:In other news...

[ginbot462]

That's modded funny, unfortunately ... I see it as prophetic. I can see it now

[Cop] : We couldn't find any evidence of you stealing music. However, we do have this video of you walking while listening to your iPod. AND, you intentionally blurred youred face to hide your identity.
[Dude]: That was a news report on obesity, the local news did that.
[Cop] : Tell it to the judge fat ass.

Re:In other news...

[Pete LaGrange]

New laws aren't needed. It's already a crime to tamper with evidence.

Re:In other news...

[Punko]

If the image is altered prior to the image being declared evidence, then it is not illegal. wiping your fingerprints from a glass is not illegal, unless the authorities have already tagged the glass as evidence.

Re:In other news...

[djupedal]

"It's already a crime to tamper with evidence."

Looking like I might do something bad later (blanked out face on photo) isn't evidence that I will...not yet, at least :)

Covering up isn't enough

[noidentity]

Of course with PDF even if it appears to be completely blacked out, it might be still readable by copying and pasting the text.

The perfect setup

[Anonymous Coward]

How soon before someone Photoshops in the face of someone they don't like into KP, obscures it, then releases it knowing the cops will unobscure it and arrest him?

When he claims "it wasn't me" will he be able to prove it? Sure, the cops are supposed to have the burden of proof but when it comes to bogeyman charges like terrorism and KP juries usually go with the prosecution regardless.

Re:The perfect setup

[whyloginwhysubscribe]

Well they say that there are features other than his face that can identify him, so in this case I think it wouldn't be possible. (Hope they get the bastard)

Re:The perfect setup

[computational super]

They need evidence to convict you of crimes against children in Germany? Wow, it must be way tougher for the police there than here in America. We only require actual evidence for minor things like murder.

Re:The perfect setup

[binarymax]

The images are used to find the person. Then witnesses (the victims) are called in to ID the criminal as part of the judicial process.

Re:The perfect setup

[pclminion]

So? The victims in these cases are children. You just tell them who to say did the deed. No "deed" ever need be committed in the first place. A picture of a naked child is not child porn in and of itself. You could take some innocent bath-tub image, 'shop a Supreme Court Justice into the picture, then tell the kid you'll give him/her 500 pounds of chocolate if they say "Mr. Thomas did it."

Or you could tell them that mommy and daddy are going to go away forever if he/she doesn't say what you want them to say. There are a million ways to pull it off when kids are involved.

Re:The perfect setup

[querist]

There are ways to detect this type of alteration. One of my fellow doctoral candidates was working on it while we were at the university together. I graduated and lost contact with him, since my research was not even remotely related to image processing.

What I do remember is that when he explained it, it was quite impressive and very, very difficult to circumvent.

Short story: Think of the ways people can ID those faked 9/11 pictures where the guy was supposedly on a balcony when the plane hit. Now, apply some of those same analyses mathematically to the data in the image file rather than simply observing the image produced by rendering the data. It involves fractal analysis and comparisons between the different elements of the image in order to identify the parts that were added to the image.

Then again, I wonder how many law enforcement agencies are aware of this technique. I hope enough will be so that innocent people are not jailed.

I hope they catch the RIGHT guy and put him away for a very long time. Child molesters do not last very long in US prisons by what I've heard.

Re:The perfect setup

[uncqual]

Perhaps it's already been done - and we are looking at the result.

Maybe the police will get 100 reports of "hey, that's Mr. Smith, the third grade teacher down at my kid's school" so they pick Smith up and make his life hell for six months... ...until they can't find any collaborating evidence and realize Chester is really much smarter than they thought. Because, Chester is a really smart image processing expert who has developed breakthrough algorithms to use pictures of Smith gathered from public sites (using insecure wireless etc. to maintain anonymity) from multiple angles, built a 3d model of both Smith's and his own head and then to locate his head in every pic/frame and substitute Smith's head for his (making alterations to account for angle, lighting, other objects etc,). Oh, and then twirling the result to amuse himself watching the police salivate over their own brilliance and buy some time to slip away while police are focusing on Mr. Smith.

Although, given the misinformation on this thread, I think it's safe to say that Chester isn't on /. (or is he and he expects the police will therefore rule him out as insufficiently clever to pull this off even if they start to get close to him?). However, maybe all the top image processing experts should be put on the "possible suspect" list.

In the unlikely case that Chester gets the last laugh on this one, let's hope it's really his last laugh because he dies laughing (and when his mom is going through his stuff in the basement, she finds and reports his stash of originals so there is some closure).

Hmm

[SamP2]

I wonder when I'll be able to buy the software that automatically unscrambles all the pixelated regions on my rather specific-content Japanese DVDs.

Re:Hmm

[yoprst]

These are not pixelated regions. That's how Japanese gals/guys actually look down there...

Re:Hmm

[EnsilZah]

Takes balls, making a porn joke on a child molestation thread.
I salute you sir.

Comment removed

[account_deleted]

Comment removed based on user account deletion

Re:I bet it's made up

[bentcd]

Look at how police are compensated. They get rewarded for closing cases. The focus is on arresting a suspect and getting a conviction. Whether or not it's the right person is not part of their pay package. Actually it is - it comes bundled with the "getting a conviction" part. If you were to demand any stricter proof than that - well, you'll need a psychic. Preferrably one that actually works. It is unclear to me how this could be the basis of any sort of sensible compensation system.

The best solution

[Dan East]

How about if they stop molesting children. Period. I doubt it makes much difference to a 4 year old whether or not photography is involved while they are being sexually assaulted.

Dan East

Re:The best solution

[finkployd]

I doubt it makes much difference to a 4 year old whether or not photography is involved while they are being sexually assaulted.

I imagine it makes it a little worse knowing that for the rest of your life sickos will be trading images of you being molested like baseballs cards.

Dalton?

[The Yuckinator]

Am I the only one who thinks this guy looks a bit like Dalton McGuinty? Given that we heard about his other habits right before the last Ontario election.....

Blurring is better than obscuring

[MikePlacid]

you might as well just opt for the classic black bar to begin with.

Black bars are so passe. Bluring/pixelating is better. Just don't blur your own face. Take your sheriff's photo and blur it all you like...

The Doctor?

[FlopEJoe]

See, I'm not a fan of the story arcs, "the Doctor is mistaken for a bad guy." For one, they never seem to go very far. Like that guy harassing Martha's mother in the middle of season three? I thought that was going to go somewhere and...

Oh wait, this isn't about Doctor Who... never mind.

Re:I bet it's made up

[petermgreen]

the thing is cops have a lot of power to fabricate or destory evidence. How is the court supposed to verify that the gun the cops claim was in the victims house and the bullet they claim was recovered from the crime scene really came from those places?

Looks pretty easy....

[gweihir]

There will be some loss at edges, but this "twirl" looks like a completely reversible operation to me. I seriously hope criminals will keep being this stupid.

Blurring is different, it removes inromation. In some cases blurring might just not remove enough information. For example (as one of the lonks in the story shows), blurring keeps the sum of black and white constant in an area. If you then have to distinguish between a sign with little black in it and one with a lot, that is still possible with blurrs. What is not possible is to reconstruct faces from blurrs. Unless you have several blurred images of the same face or a movie. Then unblurring faces could become a reasl possibility.

Fine, I will discuss the technique.

[Lethyos]

It is patently ridiculous to imply there is some secret or sophisticated method to undoing the twirl effect as available in Photoshop. All one needs to do is find the bounds of the distortion (which, admitted, is a painful process of trial and error), then perform the distortion with an opposite value to the original. This particular effect is not intended to destroy any pixels, only relocate them, so restoration is intuitive. You can all try this at home: simply load an image, twirl it, then perform the inverse as I describe. And with a little determination, you can successfully reverse the effect on the article photograph (as I did).

Blurring not that different from twirling...

[weierstrass]

As others have pointed out here, a blur, especially a gaussian blur, is mathematically trivial to undo. The only difficulties come from the fact that the blur an image editing program applies isn't a true gaussian - because of rounding and the fact that the picture plane isn't infinite.

covering may not be enough either

[Atreide]

remember some cases when people deleted data in word documents but did not purged the undo history

or cases when people added black onto a pdf document, without converting the document (open it with acrobat writer & remove the color)

as always, people must know what they do

Manhunt?

[RandoX]

Am I the only one disappointed to find out that this doesn't have anything to do with the Rockstar game?

Re:I bet it's made up

[bentcd]

the thing is cops have a lot of power to fabricate or destory evidence. If the court system does not take into account that cops will be selfish, career-seeking bastards, then there is something wrong with the court system. These characteristics must be expected of all people, always, when designing such a system. Not because all people actually /are/ like that but because enough of them are that it will become a problem if it's not provided for.

Interpol's got nothin on CSI

[Jtheletter]

If this were an episode of CSI They could have simply drag-and-dropped the photo into their "automatically un-distort face in image" program, then zoomed in over the man's shoulder to read the artist's signature of a painting behind him. Then recognizing that these paintings are only sold from one obscure store in New York City, they drag-and-drop the photo into their "compare to every frame of every NYC ATM to this picture" program and found a frame of him standing conveniently in front of his license plate, which they could also zoom in to read the registration sticker text.

Get with the times Interpol. Sheesh, CSI wouldn't even have had to use their "match a partial fingerprint I zoomed in 6000% to get off of a glass of water in a 72dpi jpeg to every known felon in the US in under 10 seconds during witty banter" program to solve this one!

Re:Interpol's got nothin on CSI

[fast turtle]

Of course not: CSI has a quantum computer that uses the spooky effect to assemble all of the pixels no matter when in time they are. That's why they're able to do so many image recoveries.

Don't forget that Las Vegas is in reality a Quantum Computer for Area51. How else do you think the government could hide the massive computer resources needed for deciphering those unknown alien languages and tech manuals?

Re:Interpol's got nothin on CSI

[LynnwoodRooster]

Dude, it's freaking Hollywood!

They've got a bigger budget than Interpol...

Re:Interpol's got nothin on CSI

[dave420]

Soon they'll just have a "solve crime" button and be done with the entire show :)

Rellying on the CSI effect

[DrYak]

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 ".

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.

Re:Rellying on the CSI effect

[nosfucious]

Regardless of the crime, a bozo that films or photographs themselves doing it deserves to be caught.

As here in "Yes, Minister"

Sir Humphrey: "He that would keep a secret, must keep it secret that he has a secret to keep."

Moron.

Interpol.

[jackpot777]

I was afraid to use the word Interpol because I thought some of you young hipsters would have started posting random lyric snippets from the band.

Re:I bet it's made up

[gl4ss]

seriously though, the cops aren't paid by how many cases they solve or get bonuses. and in this case, the method to retrieve the face is rather simple. so made up? quite probably not unless they made up the photos they found on the internet too.

point of having multiple levels in justice systems is to get to be relatively sure about things. thus, investigators are different from prosecution who are different from defense who are different from jury who are different from judge...

much better than everyone of those being just one person, like in some systems.

And why should anyone believe them ?

[S3D]

They apply reverse twirl, probably deconvolution, and some enhancement filters until they got something which looks like a picture of a man. Or they let artist to enhance the resulting image until it looks like a man. But why should anyone believe that the resulting picture is of the same man that was before the image was scrambled ? The only way to prove validity of the method it is to disclose it and let everyone to check it on several dozens of images.

Comment removed

[account_deleted]

Comment removed based on user account deletion

what i've always found crazy about digital images.

[yodleboy]

what i've always found crazy about digital images... i'm looking at a blank monitor screen (don't get smart, i have 2 monitors). on that screen, every possible image no matter how private, or bizzare is available by just activating the correct pixels. seriously, every scene from history. every fantasy. in photo realistic detail. what if there was a program that would plow through every possible combination. what images might pop up? why waste bluegene on chess? creepy.

Interpol, eh?

[BorgCopyeditor]

For their next trick, Interpol will turn dark-side-of-the-earth satellite photos into something legible using their custom-built Photoshop plugin, Turn On the Bright Lights.

Distorted not blurred

[orb_fan]

Please lets get the terminology correct - the perp's image was distorted with a swirl tool, not blurred. The distinction is that distorting moves the image data to different locations based on a mathematical formula, usually with little change to the data, run that formula in reverse regains the original image.

Blurring spreads the data from a single pixel over an undetermined number of other pixels. Depending on how the blur is applied gives varying degrees of difficulty in reversing the blur. Applying a Gaussian blur to the whole image or section usually means that the Gaussian is used once on each pixel so the new pixel's value is a known sum of the pixels around it. However, most image packages have a blur tool that blurs what's under the cursor, so the movement of the cursor determines what the contribution is to the final pixels.

It's great that the prep obviously had no understanding of the difference, as with any luck he'll be caught.

Worrying possibilities...

[rcasha2]

If interpol or some other "good guys" can do it to identify the bad guys, the bad guys could do it too. How about the many photos in which victims' faces, or vulnerable witnesses, or whistleblowers, are blurred in some similar way. I can't help imagining some poor sod who, years back, gave evidence that put some members of a dangerous gang behind bars, feeling safe in the knowledge that the only photos which appeared had their faces blurred out, then reading this news item ...

CSI

[gringer]

1. Zoom on that region there
2. Untwirl
3. Enhance
4. Over there on the right
5. Enhance
6. Enhance
7. ???
8. Profit!

Multiple Images helped Interpol

[fast turtle]

Everyone who's talking about blur/twirl/pixelation has forgoten the important thing. Interpol had multiple images. Think of this as having several frames from a film. This gave them enough information to not only un blur/twirl but to compensate for pixelization. So there was a bit more then that used to get the recovered images

holly crap! I have seen this guy!

[roman_mir]

This is what interpol site says:

INTERPOL is seeking the help of the public to try to identify this man, photographed sexually abusing children in a series of images posted on the Internet......

I have seen that guy somewhere, I remember him because of his very distinct features.

Even Easier

[Slashdot Parent]

Even easier would be to just paste someone else's face over the face you are trying to obscure. After that, you can apply your favorite distortion technique.

Result: the distorted part of the image will still look like a distorted face. But if someone manages to reverse your whiz-bang distorting technique, all they'll see is that you pasted some other face over the real face.

Re:Crucial overlooked ideas

[djh101010]

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.

Legalities of image use...

[pev]

In the practical world we live in this is great for identifying the ****. However in the legal world his lawyers live in surely all they have to do is claim that the prosecution is relying on a doctored digital image as the sole evidence...?

~Pev

Re:Legalities of image use...

[east coast]

However in the legal world his lawyers live in surely all they have to do is claim that the prosecution is relying on a doctored digital image as the sole evidence...?

Not knowing where this guy is from and not knowing how international law works (IANAL)...

In the US the restored image could be used in an attempt to get a warrant. Assuming that someone knows the guy they may turn him in and the police can file for a warrant based on the picture and the tip. Once the warrant is validated by a judge other evidence would be picked up. So it wouldn't be sole evidence in this case since more of the suspect appears in undoctored areas of the photo and any potential evidence that could be gained through further investigation.

Re:Legalities of image use...

[jackpot777]

They know when the pictures were taken, based on the file info (mentioned in the AP story).

Once they get their suspect, the authorities check his past travel details. Every travel visa, every trip made on his passport, every aircraft manifest list will corroborate the initial evidence (unless this man was sly enough to travel with forged documents - and seeing as he was cocky enough to doctor the photo in the manner he did, I'm guessing he thought he was beyond the law). And as soon as they find one photo of him with the same curtain or wallpaper pattern behind him, with or without kids, that's more evidence. Then the witnesses can be brought in.

Building a case. Gotta love it.

More Headlines Like This, Please

[Minwee]

"Interpol Unscrambles Doctored Photo In Manhunt"

"Interpol Unlocks Les Paul in Guitar Hero II"

"Interpol Tracks Down Level -1 in Super Mario Bros."

"Interpol Acquires 'Marathon Man' Achievement in Halo III"

"Interpol Microwaves Weird Ed's Hamster in Maniac Mansion"

Re:what i've always found crazy about digital imag

[Anonymous Coward]

While this at first seems like an interesting challenge, a little math will reveal the futility of it:

Even assuming only a 640x480x256 monitor, you are talking about 307200 pixels -- each of which can have 256 different colors (definately not the photo realism you are looking for).

307200 ^ 256 = 6.023 x 10 ^ 1404

This number is by a wide margin larger than the estimated number of atoms in the known universe.
(4x10^79 according to http://www.madsci.org/posts/archives/oct98/905633072.As.r.html).

Re:Crucial overlooked ideas

[sunderland56]

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....

Way easier than that....

[ProfessionalCookie]

Edit >> Undo

As if the entire system is based on good faith!

[jaypaulw]

Of course if cops want to be dodgy there is very little that can be done - especially in the short term. That pretty much holds true for anybody in a position of trust. Now obviously we want to keep positions of trust to a minimum, but they certainly seem like a necessary evil.

Next up on the agenda...

[busydoingnothing]

Interpol discovers the real secret hidden behind the Myspace angles.

Need to know the original algorithm?

[xPsi]

...or at least something about it? The twirl filter is a canned algorithm in photoshop so, as others have pointed out, it doesn't seem so shocking that one could untwirl it post hoc if just handed the image. However, given an arbitrary home grown distortion or filter, I'm not so sure it would be so straightforward. I supose if you knew the initial state was a face, perhaps that alone would help leverage some information about the mapping.

Re:Crucial overlooked ideas

[Shakrai]

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

One would hope that the "smart" criminals can think of better things to do with their talents then sexually abusing children. A "smart" criminal that manages to pull off an Oceans 11 style heist from your least favorite bank or company may become a public icon. Hell, even some of the Mafia Dons in New York became public figures and had a certain measure of respect.

But someone who abuses children? They deserve and will get nothing but contempt. Contempt from society, contempt from law enforcement ("Yeah, I had to shoot him ten times, he was resisting arrest"), and contempt from the other inmates once they are convicted and sent up the river.

This is one of the few times that I would wish prison rape on someone. Enjoy being your cell-mates bitch and worrying about that shank in the back for the rest of your life, asshole. Death is too good for you.

Re:Crucial overlooked ideas

[Anonymous Coward]

Luckily for them the moron didn't superimpose a goatse image over his face before doing the twirl.

Missing Option; Cowboy Neal !

[Qwrk]

Nea, that was a sick suggestion....... I admit. Vote me down ! ;-(

-----sig-----
"...and always remember
the last words of my grandfather,
who said '.....a truck!'.....

Counteract with a double twirl

[future assassin]

Next time you twist your photo just do it twice. Once to the left and one to the right. A black box always helps.

Anyone want to bet

[objekt]

They just twirled the image in the reverse direction?

Sometimes the simplest idea is the least obvious.

Re:Fingerprints?

[pclminion]

Interesting idea. I have a few digital cameras, at least one of them has a couple of "dead pixels" on the CCD. These pixels appear pink on the raw images. However, the effect cannot be seen if the camera is not in raw mode.

I assume the camera firmware corrects for dead pixels (probably substitutes an average of the neighboring pixels for the value of the dead pixel) so you don't see them normally. When the camera is manufactured they can test where the dead pixels are and program those locations into the firmware. But try flipping your camera to raw mode and see what you get. A particular pattern of dead pixels COULD be very strong evidence, although you still have to prove who was operating the camera when the image was taken.

Re:His photo?

[pclminion]

How do they know it's the guy and not some random someone-else? You'd have to be pretty dumb to use your own picture...oh wait, he's an idiot anyway.

Nice circular argument there. "Only an idiot would place his own face in the picture. There is a face in the picture. It must be his, because he's an idiot."

Wouldn't it be funny if they did the unscrambling and some well-known political figure's face emerged?

Re:Cue the /. legal experts in 5... 4... 3... 2...

[pclminion]

Your "cue" has brought precisely zero people out of the woodwork. The technological aspect of this story is far more interesting that the legal aspect -- I think we can all agree that sexual predation on children is horrific. Nice try. Actually, I take that back. Pathetic try.

Re:what i've always found crazy about digital imag

[pclminion]

Also interesting is the fact that a pixel-based device can only represent a FINITE number of images. At, say, 24-bit color with a resolution of 1280x1024, there are 2^24^1280^1024 possible images. That's an enormous number, but it's FINITE.

What does it mean though? Are there an infinite number of images out there that we can never see, due to the limited ability of our display devices? In actuality the human eye is limited in its ability to distinguish color and spatial variation, and so even our finite screens are probably sufficient to display anything we might ever be interested in looking at.

Data Masking != Data Removal

[gknoy]

I simply do not understand this obsession with mosaics, blurs, or twirls to obfuscate data. Why not remove the data? A big black bar of pixels is very difficult to recover from. Or, replace one's face (in the image) with Paris Holton's, or Margaret Thatcher, or any other well-known person.

OTOH, what's to say that the person didn't already do that in THIS case. If they replaced their face in the image with someone else's, and THEN did the blur/twirl, that might be a very clever form of misdirection.

Re:Data Masking != Data Removal

[WillAdams]

In my case, I blurred 'cause that's what the journal asked for in this case --- one consideration for the most recent such request was that there was medical equipment over part of the face and they didn't want obscured / detracted by a black blob --- other instances get the black block (and the original layered image is _not_ placed in the pages, only a flattened version).

I guess the question in this instance is, can one recognize a face from a ~20 x 17 pixel graphic (which is approximately how many 38 pixel^2 mosaic squares there were --- and this will be halftoned and printed, so will further degrade, and there's a large breathing tube &c. over part of the face.

William

Re:Crucial overlooked ideas

[mce]

That might be doable with one picture, but to do it multiple times is in another league. For starters he'd have to have multiple suitable pictures of his victim, fitting the pose and lighting of each of the filthy pictures that he'd want to plug them into.

Just for fun and for exploring technology, I once tried to photoshop one picture of my own head onto another of my own torso. Even though I was wearing very similar shirts (color, but especially shape of colar) in both pictures, it was bloody hard to do. And even though it looked reasonable in the end, it never fully worked. One part of the hair had an identifiable false edge, and the lighting was not perfect either (too much difference in direction between both originals).

More complicated than untwirling

[skintigh2]

Of course you need to figure out the size and center of the twirl and its angle, but I'm guessing the twirl function didn't just relocate pixels but overlapped them and averaged them. In a way, the overlapped areas would be like a one-way hash function: there would be no direct way to calculate/decrypt what the original data was, but a program could be written to guess-and-check all the possible inputs to find ones that match the output (like a hash collision) and then determine which was the most likely input.

It actually sounds like a fun program to write...

BHS - Bank Hacking System

[gr8dude]

This program was probably built with BHS SDK, the Bank Hacking System; a very advanced kit which can do many other things automagically.

Re:what i've always found crazy about digital imag

[rk]

And further, if you had some ultra-new bitchin' LCD screen with 1 microsecond refresh rate and ran through the changes at the limit of the screen, it would take 6.023e+1404 / 1e6 seconds which is about 1.9e+1385 million years, quite a bit longer than the age of the universe, and more than enough time for every star currently burning in the universe to cool to a black cinder a degree or two above absolute zero.

l33tn3zz

[billcopc]

Uber uncrackable image doctoring script:

1. Select a rectangle around your face
2. Fill with black
3. ????
4. PERVERT!!!!!!!

For everything else, there's unsharp mask. You'd be surprised the kind of things you can "enhance" from a chunky low-resolution image.

Anamorphic Art History

[searchr]

During the 18th and 19th centuries there was a brief art movement, heavy on the optical illusions. Basic mathematical gridwork would create massive smears of paint along a dozen yards of wall. But stand at one end and look down the length, the mush would coalesce into the intended landscape or portrait. One of the popular parlour techniques was to create a seemingly random "swirl" of color on a flat table/canvas. The viewer would place a mirrored or polished metal cylinder or inverted cone onto the table, and the swirl would transform into a face upon the cylinder.

In other words, this trick is SO old, that if Interpol had brought H.G. Wells in for questioning about trying to tamper with a time machine museum "mock up", and Wells saw this photoshopped photo with the face swirl sitting on the inspector's desk, he would have made a cone out of an empty paper towel roll and some tinfoil, and easily revealed the identity of the suspect.

Then he would probably have recognized the face as belonging to his old nemesis Jack the Ripper, and would then quickly dismantle his makeshift viewer and begin to look for a means of escape.

Re:Crucial overlooked ideas

[elrous0]

A guy that clever would be a shoe-in to win a "Genie" at this year's Evil Genius Awards.

Re:Crucial overlooked ideas

[diskis]

You tried once?
Professionals do that easily. Changing a lightning in a picture is not hard. If you have an evenly lit face, it is very easy to shade and highlight. Even if half of the face is obscured by darkness, it is possible to mirror the face to fill out the missing part. Not exactly rocket science.

Many sightings already

[laddiebuck]

The BBC's coverage of the story is as usual far less sensational and more informative: "The pictures had been manipulated to disguise the man's face with a swirl pattern, but computer specialists at Germany's federal police agency, the BKA, worked with Interpol's human trafficking team to produce identifiable images.". The other news (that story is from yesterday) is that there have already been quite a few reported sightings, although naturally most are probably just overreactions.

Re:Crucial overlooked ideas

[mce]

I won't say that photoshop professionals can't do better, but please note:

1. I did not say I'm a photoshop novice, I only said that I tried that particular thing once and thet the result was no perfect;

2. Mirroring faces does not work, as human faces are not symmetrical.

   Try it: take a picture of a face, and make one in which the left half is a mirror image of the right half and one in which the right half is a mirror image of the left half. Not only will you find that both versions differ, but it is even the case that one of them consistently looks more friendly/pleasant than the other one. This implies that if you just mirror a complete face, the manipulation is easily detected, since the wrong half will now yield the nicer picture.

   Besides, the whole mirroring thing can not even be considered if the picture shows the face at angle.

Re:what i've always found crazy about digital imag

[Le Marteau]

every possible image no matter how private, or bizzare is available

Yup. That's how it would work, all right: random noise... random noise... random noise... WHOA! A PIC OF WHO REALLY SHOT JFK!

Re:Crucial overlooked ideas

[Frogbert]

I can just imagine the police photo doctor if that was the case:

"And so we just un-twirl the image and...."

"What is it?"

"Damn, Rickrolled again!"

Why is the image not clearer ?

[snoddy]

Interpol must have the technical resources to do a better job than a fudged untwirl. The image, with distorted left side, is still rather vague. But, perhaps encouraging folk to take their suspicions about their neighbours more seriously, is a good "side-effect".

ROT-26 FTW!

[solaraddict]

Try double ROT-13. Twice as secure!

twirl vs blur

[m2943]

A twirl, for the most part, just moves pixels around, and you can recover the original by simply moving the pixels back.

A blur is a transformation in which information is actually lost as pixel values get averaged and quantized. When people recover information from blurred images it's based on knowledge about what was originally there; for example, if you know it's one of 10 digits in a particular font, then you can often figure out what the digit was even if the image is blurred. You can probably not recover high resolution face images from blurred images because it's the fine details that matter.

Re:His photo?

[deviceb]

or if you wanted to frame somebody.. using a simple twirl filter that can be un-twirled would be the way to go.

Re:His photo?

[Geminii]

Wouldn't it be funny if they did the unscrambling and some well-known political figure's face emerged?

You assume this is the first time they've used the technique :)