Technology behind project Natal
23, January, 2010 - Posted by mirror2image | Uncategorized
9 Comments
Sorry, the comment form is closed at this time.
About
Top Posts
Blogroll
downloads
My places
Old homepage
People
Symbian
Video
- 2d 3d 3d imaging Anathem Android arxiv Augmented Reality Augmented Reality advertisng bionic eye brain-computer interface bundle adjustment C++ camera carbon nanotubes cnn Coding Coding AR compressed sensing computer vision convnet convolutional network CPU cuda cuda-convnet Cyborg Deep Learning FAST feature detection FFT fiduciary marker fun Games Google History hologram image processing Image Registration iPhone L1 minimization machine learning Maemo Markerless Markerless tracking Math Mobile MSER N95 Neal Stephenson neural network Nintendo DSi Nokia Nokia 5800 Nokia N95 nonlinear optimization optimization pose estimation Rainbow Ends RjDj robust statistics Samsung i8510 Samsung INNOV8 scale space Science SIFT SURF Symbian Symbian Signed technological singularity Technology Tower Defense Usability Vernor Vinge videoglasses Vuzix Wrap 920AV xkcd
Twitter
Tweets by sergecellArchives
- December 2014 (2)
- November 2014 (1)
- February 2014 (1)
- January 2014 (2)
- December 2013 (1)
- November 2013 (2)
- October 2013 (2)
- September 2013 (1)
- May 2013 (1)
- January 2013 (1)
- December 2012 (1)
- March 2012 (1)
- January 2012 (1)
- November 2011 (1)
- September 2011 (1)
- August 2011 (1)
- July 2011 (1)
- May 2011 (2)
- April 2011 (5)
- November 2010 (3)
- September 2010 (1)
- April 2010 (1)
- February 2010 (4)
- January 2010 (6)
- December 2009 (4)
- November 2009 (4)
- October 2009 (7)
- September 2009 (2)
- August 2009 (4)
- July 2009 (8)
- June 2009 (5)
- May 2009 (1)
- April 2009 (3)
- March 2009 (11)
- February 2009 (10)
- January 2009 (12)
- December 2008 (10)
- November 2008 (12)
- October 2008 (10)
Meta
Mirror Image 
Have you ever tried looking at an infra-red remote control through a cellphone camera? It seems to detect the blinking pattern flashed by the remote once a button is pressed (and show it in visible light). I don’t know if that amounts to anything, just an observation.
Anyway, infrared stereo cameras equipped cellphones can also be used to read invisible IR markers which can serve many interesting uses.
Cant find out, is it related to the ToF[1] cameras? Right now I’m working with the PMD[2] Cam-driven SLAM and interested to play with markerless AR. ToF cameras looks like nice playground for AR apps (64×48 3D points, 25fps), but is not yet too compact and cheap. Here is how the data output looks: http://twitpic.com/zdi9d (yellow is PMD-captured points).
[1] http://en.wikipedia.org/wiki/Time-of-flight_camera
[2] http://www.pmdtec.com/products-services/pmdvisionr-cameras/pmdvisionr-o3/
@cfr
Absolutely not (that is if I’m correct in how it works). TOF is something a lot more technically complex, based on the capture of the phase, as I understand. Here is a simple application of computer vision technics.
The explanation in the article isn’t correct.
First of all PN have only 1 IR camera, not 2 (as you can see in the scheme posted).
The technology used by PN to determine the depth value isn’t triangulation. PN uses a pulsed IR light and a camera with an IR filter and a shutter to generate a b&W image in wich the gray level indicate the depth value. The technology is called fly time and is the same technology of the aircraft radars, submarine sonar.
Project Natal tecnology IS time of fight, i’m absolutely confident. The scheme you posted is very clear. Look: where you see the 2 IR cameras?? I see only an IR light and an IR camera.
I’m an optical engineer and I work in metrology, you may be confident that I’m right.
Thanks for clearing that rigelpd. In that case this tech is of no interest to me. Not applicable for mobiles it seems.
IMO, stereoscopic cameras would be nice, but in terms of affordability, the IR/camera concept makes practical sense. Besides, there could be future IR applications that would only need software, rather than a hardware upgrade.
BTW, I’m really lookinng forward to PN!
Anyway it is possible that Project Natal technology isn’t Time of Flight…In this case I think that the technology used by PN to determine the depth value may be reflectometry. If PN uses reflectometry the IR light isn’t pulsed and the player will need to scansionate his body every time he begin to play. This because reflectometry need to mapping the reflectivity to work well.