Become a fan of Slashdot on Facebook

 



Forgot your password?
typodupeerror
×
Technology

New Smartphone Camera Could Tell You What Things Are Made of 78

Zothecula writes How would you like to be able to know the chemical composition of something, just by taking a snapshot or video of it with your smartphone? You may eventually be able to, thanks to a compact hyperspectral imaging camera being developed at Tel Aviv University. "Hyperspectral imaging involves scanning light spectra not visible to the human eye, in order to identify the unique electromagnetic 'fingerprints' of various substances and processes. While this can already be done with larger cameras, a team led by Tel Aviv's Prof. David Mendlovic is developing a much smaller optical component that could conceivably be built into a smartphone. It utilizes MEMS (microelectromechanical systems) technology, and is reportedly 'suitable for mass production and compatible with standard smartphone camera designs.'"
This discussion has been archived. No new comments can be posted.

New Smartphone Camera Could Tell You What Things Are Made of

Comments Filter:
  • Is that where we're headed?

    • Well yes.
      Wouldn't that be great.
      Infrared camera, to detect if the heart is beating or not. Being able to see what things are made out of.
      The thing Star Trek missed, was the fact that they had horrible screens, and are really bulky, and they never seemed to play flappy bird on it.
       

      • And with their pads, the concept that a pad could hold more than one document at a time. You'd see someone who "had a lot of work to do" carrying a dozen pads or have them strewn about his desk.

        • It was undoubtedly done for artistic effect, but consider people who use multiple monitors instead of constantly switching what's displayed on a single one.
          • Of course, but I'm talking about cases where the assignments are being handed to them one pad at a time. The pads are being used at that moment for storage, not display.

        • I figured it was a military thing. Most of the time the CO reads the data and marks that he read it, and gives it back down the chain. Having it on separate devices probably adds to security.

    • https://en.wikipedia.org/wiki/... [wikipedia.org]

      Sure, this "real" hyperspectral imager can't do VISOR's "1 Hz to 100 PHz" (per the above), but it's still impressive nonetheless--assuming it's not vaporware, of course.

  • I saw an academic talk on MEMS applied to mass spectrometry about a decade or so ago. It was promising to be the next greatest sensor; smaller, faster, lower power and sample requirements, with better resolution across wide mass ranges. For the most part it hasn't made it yet. It has turned out some interesting data in a few labs but it is far from being commercially viable.

    Hopefully this group has figured out some other way to make it work for this specific application, and someone can build on it from there. It is impressive technology.
  • by Anonymous Coward

    Not to be used on chicken McNuggets.

  • Now we just need one those software that generates a full HD image from 100x zoomed 186 pixels pictures.
  • What is this a reference to? (and yes, it is on-topic if you know the answer)

    http://travelinlibrarian.info/... [travelinlibrarian.info]

  • So the question is, "Will it detect saline or silicone under flesh?"
    • So the question is, "Will it detect saline or silicone under flesh?"

      My guess is it won't be long after this tech is made available to the public that some hackers will write a filter to give us the x-ray goggles I read about it the back of old comic books, for real...

      (Note: I mean hackers in the oldschool use of the term)

      I'm not a light expert by any stretch, but from what I understand there are wavelengths of light that the human eye cannot see (but potentially these new cameras could; they already see infrared) that can penetrate substances that human-visible light cannot

  • by 93 Escort Wagon ( 326346 ) on Monday April 06, 2015 @10:26AM (#49414491)

    If I want to know the chemical makeup of anything, I'll just ask Moss to smell it for me.

  • by Fortran IV ( 737299 ) on Monday April 06, 2015 @10:30AM (#49414525) Journal
    Atoms. [xkcd.com]
    • From TFS:

      How would you like to be able to know the chemical composition of something, just by taking a snapshot or video of it with your smartphone?

      Or, more simply, I can tell things are made of "stuff" just by looking at them. Not sure why a casual user with a cell phone would care about anything more specific.

  • Nice for jewelry (Score:4, Insightful)

    by paulpach ( 798828 ) on Monday April 06, 2015 @10:41AM (#49414607)

    Would come in handy to see if that piece of jewelry is really gold and the diamond is not just cubic zirconia

    Of course, it could be fooled by gold coating, bit it is still better than nothing.

    • Re:Nice for jewelry (Score:5, Interesting)

      by Rei ( 128717 ) on Monday April 06, 2015 @11:08AM (#49414851) Homepage

      Nice for a lot of things.

        * I'm somewhat of a rockhound. The ability to ID an unfamiliar mineral would be great.
        * I like mushroom hunting. The problem is that one has to ignore the vast majority of mushrooms because there's just so vastly many species and some can only be distinguished microscopically. But the list of dangerous mycotoxins is actually surprisingly short [wikipedia.org]. The ability to accurately detect whether there's a dangerous amount of a mycotoxin in a sample would help alleviate a lot of doubts. Those who like mushrooms for their recreational uses rather than cullinary uses would also like to know how high the psilocybin levels are.
        * The same applies to usefulness for identifying dangerous or useful compounds in unfamiliar plant species

      Hypochondriacs would kill for a device like this, if it worked - aka checking for various contaminants on every product they buy and everything they put into their body. I'm sure environmental groups would love to just be able to walk around the outskirts of a chemical facility that they're protesting and take and analyze samples with their phones rather than having to send them at significant cost into the lab. Plant growers breeders could check to see whether their fruits, vegetables, etc have a unique nutritional / etc profile, or how that changes depending on climactic conditions, soil, and fertilization practices - again without taking huge numbers of samples. Vegetarians could check for chemicals only found in meat in their food. And on and on.

      Really, there's no shortage of things one could do if they could ID chemical compositions without the effort and cost of laboratory testing today. So long as the system works well enough, that is. I have a suspicion that even if such a thing hits the market, it's probably going to only be able to pick out really pronounced spectral signatures - one might not have so much luck at detecting say an arbitrary protein at ppm quantities.

      Still, even a very limited system would be useful.

      • by Anonymous Coward

        The problem with spectral imaging is that you need to build up a model library as the spectra are just levels of reflectance or transmittance in particular wavelengths. Identifying unknown substances hinges upon there being models of like substances in the database. Having a "camera" in the use of consumers would be the first step, coming up with a way to have the community at large produce quality models will be one of the primary challenges. Identifying fringe objects will hinge on those models.

  • Without reading TFA I can assert that this area of work is not going to let the consumer unambiguously determine the composition of anything. Right now the chemical analytical techniques to figure out what the composition of a substance require huge machines and significant training in spectroscopy and there's no way of miniaturizing those techniques and automating the interpretation of the data. I'm sure you can get some information but not enough to say anything for-sure :-)
    • Right now the chemical analytical techniques to figure out what the composition of a substance require huge machines and significant training in spectroscopy and there's no way of miniaturizing those techniques and automating the interpretation of the data.

      "No way"? There are lots of things that used to require huge machines and significant training that don't anymore. While I don't think we're going to see a mass market pocket spectroscope in the next few years, but I would never say it cannot be done just because we can't do it yet. I've seen cell phones used for infrared imaging already. I see no reason why they couldn't perform some rudimentary spectroscopy tasks. Sure it won't match the professional lab equipment but I doubt anyone expects it to.

      • The problem is that the spectroscopic techniques capable amenable to implemention on a small device can only give some general information about a material or mixture. This may be enough to select one option from a narrow set of possibilities (eg: is that drain pipe PVC or ABS plastic?). However, it is not going to identify the presence of a toxin in a bowl of soup or tell you that your gold watch is only gold-plated.
        • by sjbe ( 173966 ) on Monday April 06, 2015 @12:21PM (#49415625)

          The problem is that the spectroscopic techniques capable amenable to implemention on a small device can only give some general information about a material or mixture

          It doesn't have to do everything to be super useful. There is this tendency by a lot of people here on slashdot to think that something has to be a perfect replacement for existing technology to be viable. I could use something that could tell the difference between PVC and nylon right now. If it could do more that would be nice but even basic uses could be hugely beneficial.

          However, it is not going to identify the presence of a toxin in a bowl of soup or tell you that your gold watch is only gold-plated.

          That is a limitation but it's like saying that the camera in my cellphone is useless because it doesn't have a telephoto lens. There are a huge number of applications for a rudimentary spectroscope. I could use one in my factory today to check wire jacketing composition or conductor composition. (Brass or Bronze-Phosphor or Copper, etc)

          • by e3m4n ( 947977 )

            or knowing you really did get a good price on plenum cable and not some cheap pvc knockoff. Agree 100% with your assessment that despite many limitations, it will have an abundance of cheap, useful applications. Considering some of the 3rd world uses smart phones are playing in the medical field, this could only enhance the usefulness of doctors without borders.

        • The problem is that the spectroscopic techniques capable amenable to implemention on a small device can only give some general information about a material or mixture. This may be enough to select one option from a narrow set of possibilities (eg: is that drain pipe PVC or ABS plastic?). However, it is not going to identify the presence of a toxin in a bowl of soup or tell you that your gold watch is only gold-plated.

          I fully expect one of the first uses would to have a database of common toxins and their signatures, and would indeed be able to detect a sufficient sample, the real question will be sensitivity, which may limit the number that can be usefully detected until the technology is inevitably improved. And once this is in the hands of the public and accessible to develop this is the sort of thing that I expect to make some quick initial advances as others build better databases once there's a need for them, and w

          • Sensitivity is a factor but selectivity is much more crucial to any such analysis. Without physically separating the constituent materials as part of an analysis, "signatures" will overlap and obscure each other. Anyhow, I'm not pooh-poohing this technology, just trying to add some balance to the conversation, which (around here) trends towards science fiction more than science.
  • by geogob ( 569250 ) on Monday April 06, 2015 @10:49AM (#49414697)

    Hyperspectral imaging devices based on MEMS are not new. I've seen the first practical devices presented in conferences about 10 years ago. But I doupt we'll see flexible hyperspectral imaging devices like mobile telephones. Even for bigger hand-held spezialized devices it is far fetched. The problem is simply the data quantity. I deal with this problem daily with hyperspectral imaging; you are producing GB/s of data... Processing that near realtime is a true challenge.

    What I can see possible with the state of the art are devices that are specialized to identify specific compounds. But a device that can do a generic and nonspecialized retrieval and identification of chemical compounds requires a lot of processing power - especially when dealing with hyperspectral data in contrast to simple spectral data. If you want to do a quantitative analysis, its even worse.

    • by Arkh89 ( 2870391 )

      That is why people want to use compressed sensing for this. Then you never have to deal with the full bandwidth, as you're directly acquiring compressed data. Then, you also do the inference in the compressed domain.

    • by Rei ( 128717 )

      Out of curiosity, why would it be producing GB/s of data? First off, I don't even see why there would be a "per second" component in that, wouldn't there be a fixed data size per scan based on the spectral resolution? And even if that was too much the chip could pre-process that into a compressed form - while there's a lot of "noise" in some regions, spectrums also have huge regions of constant or near-constant slope.

      Where does the GB/s come from?

      BTW, there are now dozens of cell phones that record 4k video

      • by geogob ( 569250 )

        Its coming from my personal experience designing hyperspectral imagers capable of doing analysis such as what is presented in the example.

        The compression is not really the problem. The biggest issue is data analysis and interpretation.

        And the time component is obvious... you don't wand to look at the same object without moving with your handy for 2 hours to get the required SNR to be able to do a spectral analysis. You have minimum requirements in resolutions, spectral band and SNR. Theses parameter will va

        • by Rei ( 128717 )

          Okay, that doesn't really answer the question (not doubting your personal experience here). The question was:

          " why would it be producing GB/s of data?"

          And the time component is obvious... you don't wand to look at the same object without moving with your handy for 2 hours to get the required SNR to be able to do a spectral analysis.

          Again, I don't see how that explains a "per second" aspect. If you need X bytes of data, you need X bytes data, why would one be sitting around for shorter or longer than necessa

        • by Rei ( 128717 )

          Or perhaps I'm misunderstanding the concept. I was picturing that it was not taking spacial information into account, that it was just recording the spectral information for whatever was in front of the sensor. But I went to the Wikipedia page for hyperspectral imaging and it sounds like it records a full spectrum for each pixel of a whole detailed image.

          Now I think I understand what's going on here and why the data recorded would be so large. But I don't think the limitations are as great as you make them

  • So if this thing is hyperspectral, and contains the infrared spectrum as well, couldn't the output from the sensor be bandpass limited to have it act as a thermal camera as well? Cause I could actually use that. There's a gap somewhere in the insulation of my house so large that I contribute significantly to global warming.
    • by geogob ( 569250 )

      You can already buy that. FLIR, for example, is doing a thermal imaging module for iphone. And that's much easier than including a hyperspectral imager in the device.

    • couldn't the output from the sensor be bandpass limited to have it act as a thermal camera as well? Cause I could actually use that.

      There already are thermal imaging cellphone cameras [flir.com] on the market today. Haven't gotten my hands on one yet but they're pretty reasonably priced.

  • Handheld materials analysis has been possible for quite some time now with XRF technology. I guess hyperspectral imaging would be preferable to handheld XRF analyzers because of the lack of any x-ray emissions, but this is just a new solution for a problem that has already been solved for years.
  • I would like to see hyperspectral capabilities added to regular cameras and then algorithmic rendering of the spectra as overlays to the visual layers. This would be useful say in smoothing skintones by blending in data from the infrared range (or similarly for lightening foliage.) The only problem there would be that focal point for different spectra is different, so it would have to be a multishot merge (like is done with HDR.)
  • Enough people (all it takes is a few) will scream in alarm if most consumer cell phones can see infra-red well enough to create "infrared-porn" by seeing through clothing that most cell phone makers won't touch it with a 10-foot pole.

    "Is that a cell phone hidden in your pocket pointed directly at my mid-section or are you just checking to see if I'm happy to meet you?"

    • As soon as the tech for cameras to take 'x-ray' images is available to appear in phones, manufacturers will be scrambling over each other to get it in.
      It will be marketed for everything aside for the x-ray imaging, and that ability will not be implemented, so they'll be 'OK'.
      Hackers will immediately make the functionality available to the public, the cat will be out of the bag, and it'll just become another fact of the world, that you can't presume to hide your physical form if you go out in public (altho

  • Knowing if a drink is roofied would be nice... Just sayin'.

    • Knowing if a drink is roofied would be nice... Just sayin'.

      I suppose. However, I'd say that if your life choices make this a question you find yourself asking regularly, you might want to think about why that is... Just sayin'

      • by sjbe ( 173966 )

        I suppose. However, I'd say that if your life choices make this a question you find yourself asking regularly, you might want to think about why that is... Just sayin'

        Lovely example of blaming the victim. People go to bars because they like to drink and socialize. Nothing wrong with that and people aren't making poor life choices because of it.

        • I suppose. However, I'd say that if your life choices make this a question you find yourself asking regularly, you might want to think about why that is... Just sayin'

          Lovely example of blaming the victim. People go to bars because they like to drink and socialize. Nothing wrong with that and people aren't making poor life choices because of it.

          Do a lot of the bars you go to put roofies in your drinks? If so... you might want to consider frequenting a higher caliber of establishment. Note that this doesn't make the victim not a victim, or the perpetrator any less wrong. It just means that the victim should consider making safer choices.

          You don't have to rely on society/technology to keep you safe.

  • "Where No Smartphone Has Gone Before

    Wednesday, April 01, 2015 1:30:00 AM

    TAU researchers move Star Trek's fictional "Tricorder" into the real world" ...yeah.

  • Some time ago i saw some students measuring air pollution by doing a spectrogram. It could had been great to be able to scan all the photos on the internet for that kind of information, a good percent of which have associated time and location where they were taken, but you can't get that information from what normal cameras output. Maybe with this new camera, if get popular enough, the information gathered could be useful to do that kind of global measurements.
  • There's some older data on hyperspectral imaging at https://www.schneier.com/blog/... [schneier.com]

    General limitations.
    It's only skin deep. It's not a tricorder.
    Composite and mixed signals, aren't labeled for you. While hyperspectral imaging gives more-unique material signatures than RGB imaging, figuring out the most likely combination of known pure signatures, to match a noisy input signal, is still hard. Have fun with linear algebra and matrix inversion ?

Every nonzero finite dimensional inner product space has an orthonormal basis. It makes sense, when you don't think about it.

Working...