Shop OBEX P1 Docs P2 Docs Learn Events
An example of the Parallax effect... — Parallax Forums

An example of the Parallax effect...

Beau SchwabeBeau Schwabe Posts: 6,568
edited 2009-12-02 12:08 in General Discussion
As a mentor to a local robotics club that meets on Saturdays, one of the goals brought up was to design a robot with the ability to 'see' and judge distance without using sonar or IR, because it could potentially interfere with one of the competing robots.

Attached is a rough draft proof of concept that I thought I would also post to the forum. A friend of mine, who is Chief engineer and programmer for Biomedical Image Processing, 3D Visualization, and Molecular Imaging at the University of Oklahoma Health Sciences Center suggests that this method of using “stationary depth perception” is a very similar approach to how the MRI and CT imaging algorithms process an image in his line of work.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Beau Schwabe

IC Layout Engineer
Parallax, Inc.

Comments

  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2009-12-02 06:14
    LOL! When I first read your title, I thought it related to another thread in this forum — someting like the "Slashdot Effect": Person posts on Parallax about cheap dev systems in wristwatches. Forum readers stampede to purchase them, crashing the vendor's server.

    Regarding depth perception, I've long thought that a binocular vision system with a way of cross-correlating portions of the two images, could provide excellent depth information. In fact it's something I want to try with a one-dimensional camera (i.e. TSL1401-DB), using the Propeller Backpack as the processor. The missing element is a transition board to mount two of the detectors on, so their axes are collinear. The idea might work best with slit lenses, rather than something focusable, in order to blend the vertical gray levels together.

    -Phil
  • Beau SchwabeBeau Schwabe Posts: 6,568
    edited 2009-12-02 06:40
    Phil Pilgrim (PhiPi),

    "I've long thought that a binocular vision system with a way of cross-correlating portions of the two images, could provide excellent depth information." - I'd love to see something come out of this... "Parallax with Parallax Inc."

    Based on my proof of concept, you could use only one camera and achieve the 'angled' view by way of a few optics external to the camera lens to create the 'split screen' for later comparisons. That way the detection axis with respect to the camera would be collinear.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Beau Schwabe

    IC Layout Engineer
    Parallax, Inc.
  • LeonLeon Posts: 7,620
    edited 2009-12-02 12:08
    Many years ago I remember someone from MIT giving a lecture at Leeds University about a project of theirs which used standard CCTV video cameras to determine the positions of cars, people, etc.

    Leon

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Amateur radio callsign: G1HSM
Sign In or Register to comment.