SEARCH

How-To Geek

Stop Believing TV’s Lies: The Real Truth About "Enhancing" Images

sshot-112

You’ve seen it over and over. The FBI uses their advanced technology to “enhance” a blurry image, and find a villain’s face in the worst possible footage. Well, How-To Geek is calling their bluff. Read on to see why.

It’s one of the most common tropes in television and movies, but is there any possibility a government agency could really have the technology to find faces where there are only blurry pixels? We’ll make the argument that not only is it impossible with current technology, but it is very unlikely to ever be a technology we’ll ever see. Stick around to see us put this trope under the lenses of science and technology, and prove it wrong once and for all.

How Imaging and Light Prove All Pictures are Limited

4775081581_852d17ab15_o

All imaging technologies, either digital or analog, all work roughly the same way. Let’s think about cameras for a moment. All cameras create some kind of image when light (particles we call photons) interact with some sort of image creating media. In digital cameras, it is a photoelectric sensor. In film cameras, it is a chemically treated, light sensitive strip of film.

sshot-109

It may surprise you to know that film-based cameras can capture more detail than even extremely high resolution digital cameras. But even with a film camera, only a limited amount of light can be recorded on the film. The same is true with any imaging device, be it a video recorder, a digital camera, or a flatbed scanner. And since any picture is taken in a finite period of time (usually fractions of a second, in the case of cameras), there is necessarily an upper limit to the detail of any captured image.

In digital imaging, that upper limit often has to do with the ceiling the camera or device has—the number of pixels the sensors inside the camera are capable of detecting, for instance. This is all about the limits of the device itself, and is slightly different than the problem of a finite amount of light reaching the media in the camera. Put simply, no camera, no matter how advanced, has an infinite capacity for resolution.

All Data is a Product of Other Data–Garbage In, Garbage Out

sshot-113

Computers are interesting machines, but they are not without their limitations. One of things most people misunderstand about computers is that they aren’t really capable of creating “new” information, they just sort of create “different” information. In mathematics, when one part of an equation is defendant on another part, it’s called a function. When Y = X+1, Y is a function of X. Whatever, X is, Y is directly correlated.

Computers operate in a similar manner. You can give a computer a huge text file of random letters and a dictionary, and tell it to arrange those limited set of letters into words from the dictionary. This works because the end product can be broken down into a function of the set of random letters, the words from the dictionary, and the directions to create one from another.

Imagine you’re doing algebra homework on your computer. You plug in a series of numbers into your “Y = X+1” equation. First, X = 1, so 1 + 1 = 2. But what would happen if you pushed the wrong keys, and input the wrong numbers? Would you still get the correct answer? If you meant to say X = 1, but typed X = 11, would the computer still give you the correct answer? The question is, of course, preposterous. This is the concept of “Garbage In, Garbage Out.” In other words, the wrong data will give the wrong answer.

Like our equation, “enhanced” images are a function of the original image. When you start with a blurry or pixelated image (or even a sharp clean one, for that matter) no amount of filters or computer magic can coax information out of a place where that the information simply doesn’t exist. Just as “1 + 11” will never result in “2,” a limited image will never result in the so-called “enhanced” version.

Why There’s No Function to Create Data from Nothing

You might ask the question, “Isn’t is possible to create a function that can add detail to a bad image?” Well, we’re not likely to create one anytime soon. Simply because we recognize an arrangement of pixels as a face does not mean that it’s an actual face. The face part is our perception of that data—we are in fact only looking at data! To take image data and transform it into “better” data is an impossibility. A function that creates something as specific as a human face from nonsense data would require actual knowledge of the end product—you would need to know the actual person’s face in order to “find” it in the blurry image, which sort of defeats the point of this imaginary technology anyway.

It may be possible to create some kind of face-like image from garbage image data, but this doesn’t mean that that product will be relevant. It might create a face that doesn’t actually look anything like the person that was actually there. It would more likely just create a mass of pixels that sort of just looks like a “different” version of what’s there. In TV logic, there’s a face locked behind that image, and the good guys are simply going to find a way to get to it. In reality, it’s only data—and any function that recreates the circumstances of a photo being shot already has that information within it.

How to Know The Government Secretly Isn’t Doing this Impossible Thing

hs-2004-04-a-print

Lots of money is being spent by government agencies like NASA to search the sky with satellite telescopes like the Hubble and the Kepler. These scopes and others on earth provide amazing, deep space digital photography of light, and also other wavelengths in the electromagnetic spectrum—things like radio and micro waves, and high frequency radiation, like gamma and x-rays. But all of these images are subject to the same limitations discussed earlier. They are snapshots in time. A limited imaging of X-rays is the same as a limited imaging of visible light. If images could be “enhanced,” deep space photography would be easy for anyone and everyone to do. If you can “enhance” an image by zooming in on a face in a crowd, why not go outside, take a snapshot of the sky, and “enhance” it to see the details on the ground of Pluto? If this was possible, an image—any image—could conceivably contain all the image data in the universe.

Is Actual Useful Image Enhancement Possible?

sshot-110

Simply because the way trope-riddled writing presents image enhancement is wrong, wrong, wrong, doesn’t mean that graphics programs aren’t useful tools for this sort of problem. As long as the information is actually within the image, some sort of “enhancement” might make it easier to see. Take, for instance, this dark, shadowed image, lightened to show detail within the shadow. This type of “enhancement” is real, and available to anyone with a computer. The difference is the data is already there—we’re just looking at it a different way. Our eyes can’t see (depending on your monitor) the detail in the face on the left. But the “enhanced” version on the right shows us plenty of detail in the shadow, giving us a better picture of his face.


So the FBI most likely doesn’t have magic Photoshop powers, and you can’t take pictures of the little green men living on Pluto with your funsaver. Don’t believe everything you see on TV!

Image Credits: Harrison Ford from Firewall used without permission, assumed fair use. Light Writing by BloomsEyeView, Creative Commons. Garbage by Editor B, Creative Commons. IMG1189b by HooverStreetStudios, Creative Commons.

Eric Z Goodnight is an Illustrator and Graphics Geek who hopes to make Photoshop more accessible to How-To Geek readers. When he’s not headbanging to heavy metal or geeking out over manga, he’s often off screen printing T-Shirts.

  • Published 02/9/11

Comments (62)

  1. MIke Versteeg

    Actually you CAN enhance a still frame if it comes from a video, by interpolating several frames. The resulting resolution is higher than that of a single (full) video frame. Oops, did I just reveal a government secret?

  2. Eric Z Goodnight

    @Mike: Ha! That’s a good point, but that’s pretty different than what they’re putting on television. I also don’t know if that image is likely to give a much better image of somebody’s face.

  3. Gamba46
  4. ZenRuth

    Yes, but “enhancing” a still frame by interpolating data from other still frames from a video stream, well, that is just adding known information together–NOT creating data that was not there in the first place. If it ain’t captured data, it’s fantasy.

  5. Meir

    They can improve the picture of a known object by interpolating it with some (thousands) of know high quality pictures of the same type of object. This will indeed improve the quality of the resulting picture, but how much wil it like the original and not a conglomerate of the original with features of the pictures in store?

  6. Russell

    Interpolating from several frames in a movie certainly can ‘enhance’ an image…. but only if the object is still. I think that in most cases a person’s face would be unlikely to be motionless for long enough to be enhanced by this process.

  7. leecorp

    Hai,guys
    Think like this , when u get an blurry image u have the necessary info about the enviornment of the picture,,so by calculating the intensity of the pixel’s and other surrounding space ..we can calculate the other space .. like a animated image (likely) ..but making the perfect image is not a precise one…but then we can make a skeleton structure using that..then by the face detection algorithm…may be a chance …Noting Is Impossible.

  8. Piet Opperman

    If you are interested in this issue, you should see the movie Blow Up. It is the story of a celebrity photographer who takes some photographs of a woman as she meets her lover in a park. Later, when he develops and prints the photos, he thinks he sees a gun pointing at her from a hedge. He becomes obsessed with this, and makes repeated attempts to “enhance” the image, to be sure. Many adventures and misadventures follow, without any conclusive result.

    In the end of the movie, he is watching a group of young people playing on a tennis court, pretending to play tennis without racquets or balls. One of them hits an extravagant shot, and the ball (which does not exist) flies over the fence. They all look at the photographer. Clearly they expect him to pick up the non-ball and throw it back.

    Which he does.

  9. Luie

    It won’t add data, but to help with out of focus pictures, you could use a Plenoptic Lens (http://www.engadget.com/2010/09/23/adobe-shows-off-plenoptic-lenses-that-let-you-refocus-an-image-a/).

  10. MRivera42

    Yes, and “Blow Up” was made in…1966!!

    We’ve come a long way, baby.

  11. BlackSmp

    “Enemy of the state” is one of the worst examples of “picture enhancement” I ever seen in a movie.

  12. sVen56

    When Blow-Up was made, in 1966, was it shown on Netflix?

  13. WayneW

    Next you’re going to say warp drive isn’t feasible.

  14. Eric Z Goodnight

    @Wayne: Actually, the jury’s out on Warp drive. Space can literally be warped (as in distorted), so some physics gurus actually say that something like a “warp drive” is feasible. The idea is that the actual 3 dimensional space is distorted so that a line between two long distance points is brought closer together.

    Here’s Michio Kaku talking about it:
    http://www.popsci.com/entertainment-amp-gaming/article/2009-05/warp-speed-possiblewe-ask-string-theorist

  15. Nick Pilney

    Oh come on! All this talk of photo enhancement and even the video and no Fox Mulder OR Lone Gunmen references? Sad day :(

  16. Adrian E

    If you had a powerful computer, say the human brain, capable of interpreting the information captured by the photograph and adding missing information to predict a nearly accurate representaion of the scene from an obscured picture. Have you ever seen the black and white picture of a dalmation? Isn’t interpolation a form of enhancing? How does this differ from digital zoom? What does the software on a digital camera use to enhance the optical zoom?

  17. Benito

    Great article. I’ve always called “BS” when I see something like that on screen. The first video must have been made in the earlier part of the century, otherwise it would have included several clips from Bones.

    I’m no math wiz, but the equation you use as an example (Y=X+1) would still be correct if X=11, because you never declared Y to be 2. Y is “directly correlated” to X, so Y would be 12. 12 is not “wrong”, just “different”. 2=X+1? Now that’s a different story. ;-)

  18. Justin Dunn

    Thanks for writing this.
    Now we all just need to send this article to all our friends and then to the people in charge of making Crime shows and tell them to stop with the “enhancing”.
    I saw an episode of “Numb3rs” and they took a tiny picture and “enhanced” it to find out who it was.
    The episode was “In Plain Sight” ( http://www.tv.com/numb3rs/in-plain-sight/episode/526422/summary.html ).
    It’s a great episode to watch if you want to LYAO at the “enhancing” and other things they do to solve the solve the crime though :-)

  19. Eric Z Goodnight

    @Benito: That is the point of the GIGO section. Human error doesn’t lead to the right data. Yes, there’s nothing wrong or right about 12 or 2. But when you’re expecting 2 and get 12, it’s the wrong answer.

  20. Karl

    I used to call ‘BS’ on all of that as well, but have since changed my tune. Some time last year I read an article about ‘compressed sensing’. I believe it was in Wired. The idea is that most data is not truly random. If you have a low resolution picture, the concept is to find the simplest solution to the higher resolution image required.

    I am really not sure how much they can enhance an image, but it is at least possible to some level.

  21. Hatryst

    There MIGHT be some truth to this whole ‘enhancing’ thing, but certainly not fully true…
    (BTW, is it okay to discuss and explain such ‘confidential’ matters on a public forum? Don’t get yourself into any trouble…)

  22. bob

    There are some interesting techniques being developed that might change your mind.
    Check out ‘compressed sensing’
    For example see
    “Fill in the Blanks: Using Math to Turn Lo-Res Datasets Into Hi-Res Samples”
    http://www.wired.com/magazine/2010/02/ff_algorithm/all/1

  23. eeb2

    Very nice article….thanks

  24. Eric Z Goodnight

    @Karl: What you’re talking about sounds interesting, but I doubt it could be accurate enough to convict a man. I’m sure you could reconstruct *something* from it, but as for getting enough detail in a face? Seems unlikely. I’ll read up on it, though.

  25. Karl

    Just thought I would throw out some cannon fodder. I know that it seems far fetched and it might be difficult to, as you say, convict a man with it, but if the math is good and it does what they say, I think it could be very useful.

    p.s. Here is the URL for the article if anyone can’t find it:
    http://www.wired.com/magazine/2010/02/ff_algorithm/all/1

  26. Art Kennedy

    Basically agree with the author, however to a limited degree “enhancement” can heighten subtle differences in the data to make detail more recognisable to the eye in some badly exposed or slightly out of focus images.

  27. Tyler

    So what about using computers to artificially develop a crappy polaroid like in No Way Out? Is that possible?

  28. marz

    Sometimes enhancement can be done. In some special cases like is shift (blur by shifting your camera when taking photo), but also ordinary blur (if you don’t have focused correctly). You can use for example Wiener filtration: http://www.clear.rice.edu/elec431/projects95/lords/wiener.html

  29. Eric Z Goodnight

    @Karl: Just read the article. That is actually more of a way to condense images, rather than rescue data that isn’t there. Really interesting technology, though.

  30. dave

    Stuff like this is why I tell my wife that CSI isn’t a crime show, it’s a science fiction show.

  31. peko

    MIke Versteeg is RIGHT you CAN enhance the quality of object if you have the video stream.
    Such methodology named super–resolution (http://en.wikipedia.org/wiki/Super-resolution). Anybody can freely download various implementation of such algorithms and try it by himself.

    “The myth about image quality enhance” is itself a myth in the first place.

    In the movies the possibilities of course embellished, but it is a movies!
    Yell at each corner, not understanding the topic – the height of stupidity.

  32. Martin Cohn

    Focus, yes. Acclaim software made a product 8 years ago that reduced motion blur. Exposure, yes. Any number of photo editors can take a picture that appears black and get detail back. Contrast, you bet. HDR programs from HDRSoft and Mediachance can tone map or use multiple images to interpolate both over and underexposed areas of a picture.

    But resolution, nope. If you have a picture of a person that looks like a crude lego sculpture, you aren’t going to make it into a portrait. CSI-ification even strikes law-enforcement agencies. I had an Assistant Special Agent in Charge (ASAC) get angry because I couldn’t “sharpen” a license plate that was all of 10 x 15 pixels wide. He’d seen a sharpen command in photo editor software once and thought I wasn’t much of an expert. I sharpened the plate in Photoshop so the edges were hard enough to scratch glass. The plate was still illegible, but the boss was happy to have “added value” to the investigation.

  33. Fred

    The purpose of crime shows on TV is to train the public to believe that all crimes can be solved, though a few don’t get solved. Welcome to enter-trainment. The truth is that most crimes go un-solved. No, the images cannot be “digitally enhanced”, The crime scene is probably filled with too much junk to have any really good clues, and law enforcement is too understaffed to really look into everything. Ironically, many juries find defendants innocent from lack of CSI-style evidence at trial. The training doesn’t always have the desired social effect.

  34. Ryan

    TV shows do the same thing with encryption. They act like government agencies have special tools that can crack anything within minutes.

  35. Karl

    @Eric: I believe they were giving other possible uses. Look at their example with the baby and the MRI. The whole purpose was to take a lower quality image and get more detail out of it. The data compression was because it could take lower quality data and improve it. It isn’t really compressing in the MP3 sense of the word. You don’t use any compressed sensing to create the image, you actually take a lower quality image (for the camera example). With compressed sensing, you can then add in the information that you never took or stored.

    Obviously this technology isn’t perfect and can’t make an entire image out of a few pixels, but it can be used, I think, to improve an image. Obviously, it can’t make up a missing digit between two other digits on a license plate, but it might be able take low quality digits (non-legible) and turn them into something that can be read.

  36. Eric Z Goodnight

    @Karl: Email me. I think I can explain my POV better when I have a little more time.

  37. actinium

    @Eric and Karl: here’s something which is closer to what Karl might have been alluding to. Although it is in very early stages, we do see that enhancement beyond common-sense IS possible, atleast for some kind of data to some extent
    http://blog.joerg.heber.name/2011/01/14/computer-enhanced-images-with-almost-perfect-resolution/

  38. FishB8

    One of the few real-world (not to mention open-source) image enhancement tools I’ve seen is refocus-it. It can take blurred or out-of focus text and apply filters to make it readable. It is rather impressive.

    http://refocus-it.sourceforge.net/

  39. Mike

    Wait a minute, I think this is taking the wrong approach to this idea. They are absolutely right, data cannot be created from nothing. You cannot take a picture of the sky and zoom to the surface of pluto, that would be creating data where there is none because it wasn’t originally captured. But you can take the data that you have and make an inference on what the data you are missing might have looked like.

    Photo enhancement with 100% certainty is impossible, but photo enhance with 99% or 90% is probably not. People are working all the time on how to get computers to make inferences about data, this is just another facet of that and there is no reason to believe that computers could not be made to make inferences about the parts of photos that are only partially present.

    Look at voice recognition software, autocomplete technology, security software (uses knowledge of past threats to evaluate new ones), crime/terrorism predicting tools (more new age but they are working on them constantly), etc. These things work with the data they have to make inferences about what you were really trying to get at all the time. We don’t implicitly trust them because they are not 100% sure of anything but over time they can get better and better about making inferences from the data they get. Photo enhancement is just the same problem with a different kind of data.

  40. Bob_WA

    The purpose of picture enhancement in movies is to get script writers out of trouble when there is no more plausible way to solve the crime within twenty minutes or so of viewer attention span.
    The problem this creates in real life is that lawyers and juries get inflated expectations of what is technically possible, and are willing to believe “expert witnesses” who will tell them what they want to hear and show them what they want to see. Extra pixels are “inferred” by various techniques which sharpen an image, but which blur the distinction between evidence with is fact and evidence which is made up.

  41. DerrickWright

    What I want to know is how they could show that montage of cop shows where they “enhance” images without showing the clip from SuperTroopers.

    http://www.youtube.com/watch?v=v8w3fhYy6w4

  42. Hank

    @Russell: One can enhance resolution from multiple frames sourced from a video or film as you say, but the object CAN be moving.

    For example, have you ever watched a television show where they “blur” a license plate, but as the car moves, you can almost make out the number? Though while the car was still, the plate was completely unreadable? I have many times. The same goes for pixellated or blurred faces to hide identities.

    Often in those cases the camera was in a fixed position and the objects moved.

    But even if there is motion in both, so long as you can construct a motion compensation model, you may be able to improve the image.

  43. julian

    Yeah the lamest one I ever saw was CSI ‘enhancing’ a reflection in the *eye* of a person captured on CCTV.

  44. Nick

    “Call Northside 777 ” 1948 staring James Stewart.

    Good film but those who have seen it will know why I’m thinking about it.

  45. Skeev

    If you have been reading stuff online recently you may have read articles from the writers of such shows as CSI who delibertaly write the ENHANCE type sequences into plots to out od each other.
    Apparently its a gag beteween the writers and effect artists to get these outrageous picture sooving crimes throught the system.
    If you ahve seen an episode of CASTLE where he askes for the image to be blown up and then when getting it says..its all blurry you can see nothing……..So as much as this article is trying to be serious and make us take this kid of entertasinment seriously – take it from the other angle, flip it, enhamce it and then you will find its a big gag on the dim watchers who dont get it – and a in joke on the ones that know better better – that is anyone who knows how toi use a computer beyond sending emails and checking facebook

  46. Blizarre

    Hi !

    The whole point of this article is to say : if the information isn’t there, you can’t find it !
    So whatever buzzword technology you may use will either extract more useful information from the data, or just “invent” it.
    debluring, refocusing/etc. correspond to in the first case : you estimate the degradation of the image, and then deduce some sort of “inverse” transformation that will remove the degradation.
    Hyper-resolution will use the data from several images to improve the result, but this is far from enough to generate CSI-quality enhancing from regular camera !
    Compress sensing is quite off topic here.. (A MRI cannot be compared AT ALL with usual camera). And it gives an improvement only in _very_ specific cases. The math behind it are quite rigid.
    If you accepts to create information, then you can adjust almost anobody’s face on a low resolution footage…. It works, but that’s not really the best for CSI…

  47. Chaun

    Although we enjoy watching NCIS, NCIS LA, Criminal Minds, Bones, etc, my husband and I are always amused by the flying fingers typing on the keyboards as within seconds windows cascade, encryption is broken, images are enhanced, and every computer in cyberspace is instantly accessible.

  48. Jimmy D

    @ Skeev.

    In response to this being a “gag” on the dim watchers, and a “joke” on the ones that know better…. and also in response to your comment of “that is anyone who knows how to use a computer beyond sending emails….”

    It baffles me how ignorant or short sighted the human mind can be.

    Yes it might be true that “the current” camera’s are limited in the information they provide and “the current” cameras will probably never be able to provide information to allow for a good enhancment….. but for those who actually do use their “computer” and search for knowledge or to study the advancements being made every day because of computers would soon realize that, like those who said we would never fly or those who said we would never touch the moon can see that someone will create a cheap camera to take cameras in a way that the compressed sensing algorithm mentioned in the MRI can be used.

    So contrary to what the article has mentioned there will actually more than likely be something in the future that can recontruct a CSI quality image from limited information.

    So yes the shows are currently “science fiction” just as many other things were “science fiction” in the past but are now science truths……

  49. drklassen

    Actually, if you know “enough” about the imaging system that took the picture, you really can recover more information than you had. Sort of.

    See, ALL the information really is still in the picture, its just that some of the photons got into the “wrong” pixels. If you know how the imaging system distributes photons into pixels around the one is was “supposed” to put them, you can post hoc redistribute them. It’s called deconvolution and you need to know the imaging systems “point-spread function”. The better you know it, the more you can fix the image.

    This was done with some success on pre-corrective-lens Hubble images. Because they knew exactly what went wrong, they were able to fix most images.

    Of course, there are limits, and the greater the variations in the actual original scene, the harder, and more uncertain, the recovery.

  50. sheldon robidoux

    Embellished is mild word for what’s done in TV and movies. And to the extent they carry it, the programs are less real world. That is especially true of the time to go from chalk board to arrest warrant as in Numb3rs. We tolerate exaggerations in order for the storyline to become semi-plausible. And it’s best that criminals don’t precisely know what the capabilities are. The fact is that enhancement of images can be performed, within limits. But the limits are constantly moving. In many situations, it’s about removing noise and distortion. Maybe an enhanced image is not a smoking gun, but if it narrows down the suspects or gets an investigation going more the right direction, then it has value.
    In some cases the methods are tailored to suit how our brains process things. I once saw a badly pixel-ated picture in which Lincoln was totally unrecognizable. The information was still in there but needed to be viewed differently. The topic of spatial filtering in a book titled simply “Optics” by Hecht Zajac is good reading. I suspect the field has advanced much since that was published.
    The computer allows for some of this manipulation to occur in much less time than it would take to construct optics to do the same and allows things to be done that may not ever have been managed otherwise. We’ll have to wait a few decades to find out what the NSA and others were really capable of today.
    “You are hacking into the Pentagon mainframe. Press OK to continue.”

  51. Bodsda

    I think a lot of people are missing the point.

    You can only work with the data you have.

    You cannot, with a decent degree of accuracy, create the data you are missing.

    Bodsda

  52. Eric Z Goodnight

    @Bodsda: You’ve summed up this whole article in two sentences. Very zen!

  53. sheldon robidoux

    It’s not necessarily about creating data where there was none. It’s about taking information which was obscured and making it intelligible, so from a poor image or audio recording something useful is salvaged. So you can enhance fuzzy images – although not to the degree portrayed in many cases, it’s true. But the existence of such technology is not all lies. Some of it is remarkable.

  54. Greg K

    Between film hiding more info than the eye can see, and frame combination, as well as algorithms for analyzing specularity, reflections, light bending and absorbtion, one can bring out AMAZING things into the view from original film:

    http://blog.wolfram.com/2010/12/27/fixing-bad-astrophotography-using-mathematica-8-and-advanced-image-deconvolution/

    http://blog.wolfram.com/2009/03/03/secrets-of-the-universe-hiding-on-my-home-computer/

    http://reference.wolfram.com/mathematica/ref/ImageDeconvolve.html

    Enjoy!

  55. Jonathan Michael Reiter

    I don’t think it’s a lie as much as it’s the writers of such TV shows simply not knowing what they’re talking about, science or technology wise…

  56. Karl

    @Eric: Apparently I’m an idiot, I can’t find your email address.

  57. Eric Z Goodnight

    @Karl: Hardly, sir. It’s on several articles, but not on this one. ericgoodnight@howtogeek.com

    I plan to write a followup article on Compressed Sensing. I think it’s really interesting.

  58. CD Hanks

    Geez, get a grip people, it’s TELEVISION and not pretending to be science, it is entertainment. If you expect accuracy, stick with the Science Channel.

  59. CD Hanks

    Hey Fred, you said, “The purpose of crime shows on TV is to train the public to believe that all crimes can be solved, though a few don’t get solved.” Really? I thought their purpose is to entertain the viewers and get us to buy the advertiser’s products. The author of this article and many folks here are way too serious and need to lighten up. We are talking about TV shows, not educational channels.

  60. Chris

    We all know its crap, once you get to raw pixels, you can’t ‘fix’ the image as in American CSI & NSIC TV shows; come to that you can’t perform a DNA match in 15 minutes or match a fingerprint in 30 seconds.
    It just doesn’t work like that.
    Talking about work – how come the ********* don’t wear isolation clothing when investigating a crime scene, loose blond hair is nice to watch – but a bugger in a crime scene.
    Oh there’s the reason – ‘blond hair is nice to watch’………….

  61. Jon

    “Blow Up” was a great movie. I’m old enough to have seen it in a theater. For another take on photographic enhancement, I recommend the novel “Reflex” by Dick Francis.

  62. Mike J

    I believe the first commenter is correct… It is possible to do some pretty impressive enhancing with multiple frames of a person/object moving in front of the camera, or if the camera is moving. I beleive a similar idea is used for satellite imagery, and I think it may even be a similar concept to synthetic aperture radar (SAR).

    Perhaps one example would be if an object (e.g. a face) makes up an area that is only 50×50 pixels (2500 pixels). If the object moves and in the next frame the object occupies a separate area (overlapping is okay) that is also 50×50, you now have 5000 pixels that captured the object (even if the object overlaps from frame-to-frame, it’s still a different area of the object that is being picked up). Knowing the spacial relationship between the object in one frame vs. the same object in another frame, it is possible to enhance the resolution of the subject in the image to be approximately the same quality as if 5000 pixels made up the object (approximately 70×70). But if you have the object in 100 different frames in slightly different positions in each frame, it is possible to get a much higher resolution single image. Why we do not see this in use more often by common law enforcement agencies? I don’t know. As mike versteeg pointed out, the first commenter, I’m pretty sure big brother, i.e. CIA/NSA does have this capability…

    Consider this theory for creating higher resolution video (not just a single still) from low resolution video… 10 low-resolution frames are analyzed and converted to 1 high resolution frame, e.g. suppose the LR frames 1-10 become frame 1 of the HR video. Then LR frames 2-11 become frame 2 of the HR video, then 3-12 LR frames become frame HR frame 3, LR frames 4-13 become HR frame 4, etc. etc… this theory is discussed in a few of the links below.

    http://en.wikipedia.org/wiki/Super-resolution
    http://webpages.iust.ac.ir/mozayani/Papers-pdf/IPCV09.pdf
    http://www.seanborman.com/publications/SRreview.pdf
    http://www.infognition.com/articles/what_is_super_resolution.html

    One last thing to think about is regarding compression. I beleive that if a codec is being used for the video that essentially only “updates” pixels if the pixel has changed value significantly (over a certain threshold) from the same pixel in the previous frame, super resolution techniques may not work as well.

Enter Your Email Here to Get Access for Free:

Go check your email!