How to see around corners
Вставка
- Опубліковано 19 бер 2012
- This video shows how scientists at the MIT Media Lab (www.media.mit.edu/) reconstruct a hidden object using scattered laser light. Future applications may include seeing in dangerous or inaccessible locations, such as inside machinery with moving parts, or in highly contaminated areas.
Read the original research: www.nature.com/ncomms/journal/...
Check out the researchers' project URL: cornar.info
And first author's web page web.mit.edu/~velten/www/
Animations by Sam Woolf: www.samwoolf.net/
Music by Jay Marsh www.jaymarshmusic.com/
Sign up for the Nature Briefing: An essential round-up of science news, opinion and analysis, free in your inbox every weekday. go.nature.com/371OcVF - Наука та технологія
Well, I actually meant "how do they get an image of the *back* of the model?" Moving the laser just focuses on different spots, and they measure the time it takes each photon to return in order to determine depth, but I can't figure out how they got a 3D image of the back
its not possible, maybe algorithms add 3d
"Enhance!"
very interesting to see
this film is brilliantly done!
Wow, what an incredibly made video presentation. Very nice!
This is one of the best experiment i´ve seen. Great work.
Dieses experiment ist eins der besten das ich je gesehen hab. Großartig,....!!!! Ich wünsche dem Team alles gute und viel Erfolg in der Zukunft. :D
Brilliant. So creative.
This is totally awesome!
Wow! Awesome!
This is incredible its literally using radiosity as a method of generating an image. I was actually talking to my physics teacher about something like this. Id imagine this could be expanded upon with adequate sensors, you could not only record the light and topology information but also measure what wavelengths were absorbed and formulate a colour image. I originally got my idea from rendering engines :p but this is amazing. The applications are mindboggling.
Imagine how much better details the computer image would been if they got the sensor down from 2 picoseconds to just 1. Probably would been 1:1.
The timing of the return hit means nothing if you don't know where it is returning from. So it's gotta be a statistical analysis of some sort - a tendancy for different parts of the object's surface to send returns at a higher rate to particular areas of the wall depending on where the laser was originally aimed. The timing is only helpful in hypothesizing one axis of that model it generates. Cool, nice work team!
This is what lessons at school could look like (hold to the media production)!
And great job, the algorithm must have been a tough piece of work.
This is so amazing. I can't believe it.
This is awesome!
Simply amazing!!! :)
Very impressive. May not be as far away as you think, Xbox Kinect is something like this but without the corner. So I can see a low-res version (few cm) being possible soon, good enough to tell if there is something big around the corner or not.
This is just great.
This is 10 years old now, and I still think it's one of the most amazing things I've ever seen.
Its up there. I wonder how much they have discovered since... I want to know more!!! Anyone got links?
Amazing invention.
Это просто офигенно! Отличная работа!
Genious and simple,
simple use of Huygens-Fresnel principle...
this is just brilliant.
Science never ceases to amaze me.
This reminds me of how it's possible to reconstruct the image on a CRT by observing the changes in background illumination through a window as the beam traces the image on the screen. Many activity LEDs also (unintentionally) function very well as data transmitters.
I see... Thanks! I personally think it's just really cool that light can work in (sort of) the same way as sound (or am I just misunderstanding it all?) and yeah, needing a still object for reference is a pain. Though I'm not really concerned about needing multiple shots, since lasers travel at the speed of light and the lasers can be fired off within, what, the blink of an eye?
Impressive to say the least!!!!!!!
your right,things are starting to change for the better.
very valid question. The only reason I know of is that laser sensors like this give you the depth information that cannot be retrieved efficiently using just cameras. That's why in many robotics applications they prefer to use laser range sensors besides cameras.
Wow. Just wow.
Man... MIT always gets all the cool sh*t. I need to go college there so I can do some cool research projects!
this is so nice.
@Aedolon Yes; although to be fair to the technology, it is a diffusive mirror (which is why your eye can't reconstruct an image from it).
MIT ROCKS!
its kind simple ,light also have a attributes of a wave like Diffraction.
so it get pass solid object / but the "wavelength" is so small unlike sound wave/ so we rarely notice it
Awesome.
Nice!
IF ONLY physics teachers explained as nicely as this video!
Very cool...
Много думал, как создать такую вещь, оказывается не все так однозначно!
у разработки большое будущее, молодцы ребята!!!
incredible.
One could probably "send out" multiple lasers differing by their wavelength simultaneously. I think this could be used to generate the approximation of the model much quicker, though it might not be trivial. Basically if you can approximate the model faster, it doesn't matter if it's still or moving. The knowledge of disance is probably quite crucial for accurate measurements.
We are truly living in a wondrous age of invention and discovery!
It's important to remember that since background light is constant we can apply filters and transforms that will correct for the noise.
Great photo.
So Awesome!!!!
Excelent for the Cornner Gun. :(
They don't seem to mention here. But it could probably be scaled easily to anything you need. More cameras or just faster cameras. More lasers etc as you need them.
Definition of mirror (from dictionary . com). See meaning #4 (the meaning within the field of Optics).
Since it reflects light, it is, by definition, a mirror (in the field of optics, which is the field we are dealing with here).
This system works by scattering light off a surface in many directions. Mirrored surfaces would just send the light pulse away in one direction. In short, a normal camera would do a better job in that situation.
sure; rescue missions, firefighters, police, traffic/car, military, medical use, security camera's, mining, etc. And the're probably tons more applications.
@oarking No, it is indeed a mirror. It is what is known as a diffuse mirror, you can read up about this here en . wikipedia . org / wiki / Reflection_(physics)
yes he's doing an amazing job
Now that i think about it, this reminds me of those little orbs the people throw up in that new movie coming out, "Prometheus". Would this technique ever be able to map an entire room?
Bats' echo location seems to work just fine without water or lasers (although I will grant that everything is, indeed, better with lasers). It may not carry as far, or be quite as accurate due to air's lower density, but if a flying mouse can use it to find a moving mosquito, it's still pretty darn good.
I think the distance/accuracy sacrifice may even be applied to laser location, too, but that's kind of conjectured... more concerned with it taking multiple shots and needing a still object...
How do you extrapolate the back faces of the 3d images? My assumption that every object is a mirrored cylinder or sphere type of object, or do you find a photon pulse that bounces off the rear wall and then all the way back. That would quite the algorithm.
Joshua McCunn no idea about science but its probably the time and the photons position being calculated to get the dimensions of the object. like throwing a ball into a well and you calculated time and mass into distance and by that u got how long a well is
Mind=Blown
Thanks :) I'm just curoius as I'm currently working on a futuristic story, and would like to know in case I wanted to incorporate this technology into it.
What Sperael doesn't seem to understand is that it's a technology in early development. You probably would've said the silicon transistor was impractical back in 1954. Just think of the possibilities. Maybe if they could do they laser switch at high enough frequencies, and the processing fast enough, to render images in "almost" real time. That way they could see moving objects.
this song is super cool any way to find the artist?
Dose anyone know where the music is from? Tull likes it.
There are many ways to "see around corners", and see it with your naked eyes, without sophisticated cameras and algorithms. And where the raytracing explanation simply does not work. All you need is a reflecting surface, no laptop needed. You can try it at home, block the direct path of a laser, and still be able to see its reflection on a screen beyond the obstacle, simply by moving around.
You can subtract background lighting from the desired data by taking a "control" image.
This will make "I should shoot though this wall" easier for the next generations.
Did you read the dictionary entry for mirror? The word "mirror" has 5 different meanings. The meaning I am referring to is the meaning that is specific to the field of optics (definition #4) It is a precise technical definition of mirror and yes, in the optics definition pretty much anything that reflects light (diffuse or specular) is a mirror.
The system described in the video is an optical system, and is a creation from the field of optics, thus #4 is the appropriate definition to use.
The lights from the kinect are not lasers being scattered. I'm not going to go into detail, but they just detect if there is movement, or something obstructing them from reaching the wall.
Why not? It's a perfectly viable application. Especially if the mirror trick doesn't quite cut it. And it seems a lot simpler than setting up surveillance, and more pervasive than thermal imaging.
Would this be regarded as diffraction grating? Student in physics right now
@pomokey it's sensors are sensitive enough to read light information half trillion times per second. but each is not each own photograph. the result is not a video with trillions of frames. the camera takes an immense amount of data points and merges them into one 3D image. It's experimental, hopefully one day it's refined to creating live video. but right now it produces singular 3D images.
nice..
They record many images for each laser pulse. They do this for many laser pulses. They combine this data and do some pretty complex math. And out on the other side comes a very rough 3D model.
Science is exciting.
How can I determine the camera 's errors?
Anyone out there that could answer my question? How long would this whole process take? (From the moment the pulse is fired to the image being constructed.)
No. You only need one background image from the place where you are imaging from. And it works fine even without background subtraction because light pollution is constant. The reconstruction code will be able to define the object you were imaging anyway, just that the resolution will suffer a little.
since they are using laser light, they can look specifically for light at the same frequency as the source. That probably helps dramatically with light pollution, even without a background image.
Yes.
I think it would be nice to use it for military applications in urban areas, if S/N is large enough
It can be used for internal imaging, like an x-ray
this is cool.
And how long does it take to reconstruct an image?
Thumbs up!
first can we see the laser thingy which is fired second where can I get a camera like that but amazing plz answer my questions I will visit this site for everyday for my answer. THUMBS UP FOR U GENIUS. :)
There's a paper cited in the video. It will have all the specifics about the physical set up of the experiment. The hard part will be getting the program that reconstructs the data into a usable picture. Unless there's something in the paper I guess you could ask the guy who wrote it.
Look Up Allex Collier He is a ex-irs bigwig that realized what those people were doing and actually quit and started speaking out against it and let the Public know what is really going on and he is a genious on the subject and a must hear speaker
Your welcome! Think of it like a mobile phone camera. Some are really fast and high res. Some are slow and awkward. Lets hope the manufactures don't make it with buggy software! :D
cool
Everyone would have decided that, because everything is better with LASERS! But really, explain how echo location would work well on land without lasers? To me it seems like a sonar-like tech wouldn't work as well on land as in water since air is less dense than water, but I don't really know so much about this field and I'm curious.
Do you understand how a dictionary entry works? Most words don't have a single meaning... they have multiple meanings. The numbered definitions in a dictionary entry enumerate the different meanings for the word. For example the word "color" has for meaning #5 "vivid or distinctive quality, as of a literary work".
I selected meaning #4 for "mirror", which is the meaning used within the field of optics, which is the field related to our discussion, and thus the appropriate meaning to use.
Yes. For medical care to see in hard places in bodies, to take exray's with just a photo and in cars, to see if another car is coming around a corner
Good grief. This is MAGIC!
"Close enough!"
Now engage the target...
pretty cool yo
太酷了！
A simple radio radar can do it )?
The title of this video misuses the word "see" and "around corners". A more accurate title would be: "How to detect objects not directly in view.'
The camera is not "seeing" around a corner, the method is, effectively, using a surface as a mirror for the laser light to be reflected to and from an object to be captured by a camera. I remember a time when scientists were far more careful of the words they used.
mir·ror
[mir-er] Show IPA
noun
#1 deleted due to space limit
2.
such a surface set into a frame, attached to a handle, etc., for use in viewing oneself or as an ornament.
3.
any reflecting surface, as the surface of calm water under certain lighting conditions.
4.
Optics . a surface that is either plane, concave, or convex and that reflects rays of light.
5.
something that gives a minutely faithful representation, image, or idea of something else: Gershwin's music was a mirror of its time.
increase the rate at which you fire the grid of laser beams so you can build up images in real-time allowing you can see the movement
"How to detect objects not directly in view." alawson911 said. That is how our eye works, it 'detects' objects in front of us. Seeing is correct use of language. We 'see' with mirrors through telescopes, binoculars and periscopes. They use mirrors.
Worth watching - Teld Talk Ramesh Raskar: Imaging at a trillion frames per second
^^Truth.
Notice, that this works only because they know the optic properties of the walls and their relative position, including the angle between them. That is, this won't work if you'd like to use it to find out if there's a rabbit in a hole, mostly because the walls in the hole are not that flat. )))
Would it though? Would it not just be another surface for the light to bounce off of?
Damn, super legit, I hope it doesn't really only have 303 views!
915K
9 years late i know.
Wooooooooooooooooooooooow!!