Forgot your password?
typodupeerror
Input Devices Security Technology

Cheap Scanners Can "Fingerprint" Paper 88

Posted by timothy
from the to-go-with-the-mandated-yellow-dots dept.
carusoj writes "Researchers at Princeton University and University College London say they can identify unique information, essentially like a fingerprint, from any blank sheet of paper using any reasonably good scanner. The technique could be used to crack down on counterfeiting or even keep track of confidential documents. The researchers' paper on the finding is set to be presented at an IEEE security conference in Oakland, Calif., in May." Update: 03/10 22:43 GMT by T : J. Alex Halderman, Associate Professor of Electrical Engineering and Computer Science at the University of Michigan and one of the authors of the study, writes with more: "My group has just put up a site about the work and a copy of the full paper, and we will probably add a video later tonight."
This discussion has been archived. No new comments can be posted.

Cheap Scanners Can "Fingerprint" Paper

Comments Filter:
  • by Bandman (86149) <bandman AT gmail DOT com> on Tuesday March 10, 2009 @12:20PM (#27136205) Homepage

    Features that act like fingerprints.

    Things like fiber arrangement, etc

    • by Aladrin (926209)

      Hence the quotes around "Fingerprint".

      • Re: (Score:1, Funny)

        by Anonymous Coward

        Yes, it looks like we can both read. Unlike the first 10 comments or so that talked about putting tape on their fingers...

      • See, I though the quotes meant to retrieve fingerprints, as in "Fingerprint the suspect" or "Fingerprint the car door". The tag on the RSS feed said 'Cheap Scanners Can "Fingerprint" Paper' and for a moment, I thought me and my Visioneer were off to a wonderful new career in forensics... Crap....
        • by Bandman (86149)

          They changed the text after it went live (and I posted my comment). You aren't going crazy.

    • by Philom (24273) *

      Here is a link to the actual research paper: http://citp.princeton.edu/paper/ [princeton.edu]

  • yeah, and? (Score:5, Insightful)

    by girlintraining (1395911) on Tuesday March 10, 2009 @12:21PM (#27136215)

    Professional counterfeiters won't be deterred by this. It'll only catch the teenagers that try to print twenty dollar bills to pay for their school lunches. Much like how Photoshop won't edit files with a certain shade of green, or how ink jet printers embed a unique identifier in the yellow ink output. *shrug* It's amusing that most counterfeit money comes from Iran from a pair of printing presses that are identical to the ones used here in the United States, yet there's all this effort on trying to curb production from Joe Average. Most real threats come from sophisticated operations like that, and require a team to combat. This is nothing more than a novelty.

    • by cheftw (996831)

      Much like how Photoshop won't edit files with a certain shade of green

      Is this true? Google didn't turn up and I must admit I've never (knowingly) run afoul of it in my own experience.

      Link please?

    • It's amusing that most counterfeit money comes from Iran from a pair of printing presses that are identical to the ones used here in the United States, yet there's all this effort on trying to curb production from Joe Average.

      Funny thing about those perfect printing presses, for a while they were in North Korea. Before that China, and the Chinese probably bought them from the USSR. It's almost like they're an urban legend that springs up whenever there's a particular set of dastardly freedom-hating furrin

      • by Quantos (1327889)
        The presses are no big deal, any off set will do the job. What's supposedly hard to get is in this order.

        1. The Plates 2. The Ink 3. The Paper

        Then of course there are all of the security features that you need to find your way around.
    • Of course the professionals will continue to print. The reason they make it hard to counterfeit is to stop the casual guy in his basement and making a mess of the money supply.

      You can never stop the hardcore well financed criminals.

      • You do realize that the best way to get the US out of this credit crisis enchanted-debt-ring is to start printing dollars with no debt attached to them?

    • by inKubus (199753)

      This is much more than a novelty. It can verify with a high level of certainty that something is an original document (provided you trust the signature database). The uses in the legal profession are innumerable.

  • useless! (Score:1, Insightful)

    by Anonymous Coward

    Sorry, but this is only really useful in identifying leaks if the leaked document is either A) the original document or B) a high resolution/low contrast scan of the original document. Please note that documents are generally scanned at low resolution and high contrast to aid readability. The high contrast completely blows the background (i.e. the fingerprint) out.

    Also, the minute a document is reproduced (fax, copier, laser printer, whatever), the fingerprint is destroyed.

    • A scanner could be configured to randomly sample places on the page at high-res and store that information with the high-contrast/low res scanned file.
    • by 4D6963 (933028)
      No, I'm afraid you missed the point of this. The point isn't to identify a leak from a scan found in the wild, but rather to identify the origin of a paper that a forensics team would be in possession of, as to prove its authenticity. If you RTFA you would have read that it actually takes four scans of the document rotated by 90 degrees (so the light angle is different) to build that "fingerprint".
  • hm (Score:2, Interesting)

    by jperl (1453911)
    "A drug company like Pfizer, for example, could take fingerprints of their labels when they are shipped, and this data could be verified later by a government or company representative in order to spot fakes."

    In times people consciously order fake viagra or fake diet pills this might not help.
  • Neat But... (Score:3, Insightful)

    by _bug_ (112702) on Tuesday March 10, 2009 @12:38PM (#27136467) Journal

    This won't stop money counterfeiters from creating money. Even if you added some kind of barcode that contained the fingerprint of the paper to every bill, the overhead to scan the bill would make it worthwhile only to large bills, so the counterfeiters stick to small bills. Or they reverse the fingerprint process and print valid barcodes on the bills they counterfeit.

    But in terms of tracking objects, it's a great idea. If a document winds up in the wrong hands and the authorities recover it, they could then trace it back to its origins. Take it a step further and apply the concept to other objects. Maybe use xrays on components of a car to help ID stolen parts. Cost of implementation would make this work only with very high-end autos. Maybe something similar for weapons? Serial numbers can be filed down, but changing the unique composition of the metal would require a bit more work.

    The best thing is it works with existing items, so you don't have to force people to buy new items for the system to work.

    • Solution you propose is already sold in Czech. You buy spray can full of microdots (0.4mm) with unique hologram id. Use it on your car parts and you can prove your ownership even if car is disassembled to parts.
    • by bentcd (690786)

      This won't stop money counterfeiters from creating money. Even if you added some kind of barcode that contained the fingerprint of the paper to every bill, the overhead to scan the bill would make it worthwhile only to large bills, so the counterfeiters stick to small bills. Or they reverse the fingerprint process and print valid barcodes on the bills they counterfeit.

      Surely there's already perfectly good scanners out there for detecting fake currency. It's just that most establishments that handle money (stores, fast food joints, etc.) can't be bothered with the overhead of purchasing thousands of units to deploy with their cash registers, and also don't want to take the efficiency hit of running the money through a scanner (or just having the tiller look at it in UV or whatever) before putting in the till.

      What you could theoretically use this technology for is to trace

  • by JRHelgeson (576325) on Tuesday March 10, 2009 @12:49PM (#27136633) Homepage Journal

    HP released a palm-held page scanner that you would wipe across the paper like a squeegee. It would scan the text and assemble the entire page based upon the unique grain pattern in the paper. The market didn't understand the concept, wasn't ready for a briefcase document scanner, whatever the case was, but it failed and was withdrawn from market.

  • Fingerprints printed on paper can be used to fool high end Fingerprints scanners as well the mythbusters did that.

  • Correction (Score:4, Funny)

    by batquux (323697) on Tuesday March 10, 2009 @01:17PM (#27137063)

    The researchers' paper on the finding is set to be presented at an IEEE security conference in Oakland, Calif., in May.

    After a high resolution scan, it turns out this is not the researchers' paper after all.

  • Beyond counterfeiting, there are uses of this technology in criminal investigations.

    Say, someone sent a threatening letter to someone and then eventually murdered them.
    Later, the murderer denies having written that letter.
    The paper on which the threatening letter was written could be tied to the paper in the murderer's home using this kind of fingerprinting.

    Of course, the courts in general have to be convinced of the uniqueness of this fingerprint before this could be used.
  • This is not news, the university of applied sciences in Mannheim worked on this several years ago, and it is already implemented in their diplomas.
    Interestingly they discovered it as a side effect, while trying to cramp more data on a sheet of paper.

    This is the german page where you can test a diploma: https://zeugnis.hs-mannheim.de/ [hs-mannheim.de]
  • I'm currently developing a Java fingerprinting library ( http://sourceforge.net/projects/jfooid/ [sourceforge.net] ) and it's learned me if you want to fingerprint something, it needs a certain unique continuity. Fingerprints have that in their unique curves. Audio has it in the sound wave but I don't see how a piece of paper has that, let alone be able to distinguish a copy from an original.
  • by DrVomact (726065) on Tuesday March 10, 2009 @02:09PM (#27138005) Journal

    The more I think about this less than astonishing breakthrough, the less sense it makes to me. It seems to me that, as described, the technique is useful only in proving that a piece of paper is identical to itself. Unless you're fascinated by tautologies, this is not exactly exciting; furthermore, none of the uses cited in the article seem plausible.

    For example, how could this technique be used to detect counterfeit currency? As everyone who has ever thought of combining a 20 dollar bill and a Xerox machine knows, just copying the bill doesn't produce a convincing fake, because the mint uses special paper to print currency. Is the author of this article suggesting that we scan every bill that's printed, file the scans by serial number, then scan every bill that's spent, and compare the scan against the database? Even comparing only suspect bills seems impractical to me—besides, if the counterfeit is that good, not even the government wants to know.

    The pharmaceutical label verification is equally ludicrous. Remember, you'd have to authenticate each particular label against the database to verify it. This is nuts. You don't just rely on the label to authenticate lab-grade products—you rely on procedures that include traces, accountability, and a documented chain of custody. If we're talking aspirin, then the cost would be ludicrously out of proportion to the gain. If we're really worried, say if we're dealing with plutonium or something, then we're not going to rely on a silly label for authentication. How do we know the label isn't real, and the stuff in the container was stolen in transit, and something else substituted?

    Could we imagine a case where it would make sense to use this scanning method to verify the authenticity of a document? Say we have a very, very, important document. We want to make sure it doesn't get swapped out for a fake document that looks just like it? Aside from the question of why it would matter, I'd have to ask: which is more vulnerable to malicious tampering—a paper document or a database record?

    There might be applications to this technology, but if so, the article isn't telling us.

    • by Inda (580031)

      When was the last time you tried to photocopy (Xerox) a $20 bill? :)

      Even my cheap £40 scanner will not scan money. I thought it was clever when it automatically diplayed an anti-counterfeiting website after I tried. Any angle, even folded in half would display the page. Storing images of all paper money is probably why the drivers were so big...

  • I don't see how it helps crack down on forgery at all. It only enables you to identify a piece of paper you have previously had access to in order to scan its fibres. Then, if you encounter the same physical piece of paper again, you can repeat the scan (which takes several passes using the otherwise conventional over-the-counter scanner).

    It DOES enable you to identify a leaked document, if it comes back into your hands, but I don't see why you'd opt for paper fibre scanning over some other sort of hidde
  • The article does not clarify how exact they are. For example, there is a huge difference between only being able to identify that page A is still the original page A and being able to say that unknown page A came from Batch 12043, which according to our records was produced by X corporation, on Y Date, and sold to Z retailer on date W, using UPC code 90827452345 through 90827452356 Which they can do can dramatically alter the usefulness of the technology. I would be very surprised if they could do both.
  • So, it will work fine... until somebody spills some coffee on the paper or something. Big deal.
    • by Verdatum (1257828)
      RTFA, It reads texture, not color. The alignment of the fibers isn't changed by cofee. If I dye my thumb purple, my thumbprint is still my thumbprint. Now if you spilled coffee on it, and firmly rubbed it firmly with your palm, it might change a bit, but even then, it could calulate the area that is common with the original. any two random sheets of paper, even produced from the same paper-mill would have a value pretty darn close to zero. Anything slightly higher than that suggests a match.
  • (Arrg, Slashdot seems to have eaten my first attempt at this comment)
    A decent silicone mold of a sheet of paper would be able to pick up a sufficient level of detail as to reproduce something (like a resin cast) that could fool the scanner. A bit of experimentation could produce a substance with the physical properties of paper that could fool it...I'm thinking along the lines of a finer grained pulp with some stronger binding agents.

    It would take some cleverness and home-brew spirit to work out the tec
  • If you do something illegal, use a stolen scanner and destroy it afterwards.

  • The reviewers didn't do their homework; this technique has been around for decades.

"Only the hypocrite is really rotten to the core." -- Hannah Arendt.

Working...