I'm making a Laser Harp. The application needs the position of the hand along each of the laser beams to modulate a sound synthesis or modify the properties of a sampled instrument. I'm wondering if doing a Color Tracking with OpenCV will do.

  1. Make the beam visible with fog machine.
  2. Segmenting the image for each beam.
  3. Measure the height of the beam.
  4. PROFIT.

Other way can be using an infra-red light sensor like Infusion Systems - ReachClose v3.1

What method is more accurate and fail-proof? Any other ideas?


  • $\begingroup$ Perhaps some diagrams? $\endgroup$ – Spacey Aug 9 '12 at 14:06
  • $\begingroup$ Here is a diagram > mediafire.com/?rl279yt3s1nwc11 Sorry but can't upload files here because I'm a new user. $\endgroup$ – Ventolinmono Aug 9 '12 at 21:17
  • $\begingroup$ If you want to use a camera, could you just detect and track the hand instead? This could perhaps be done using a kinect to obtain range easily? Alternatively you could use a range sensor directly, google lead me to a tutorial explaining this setup which you may have already seen: makeprojects.com/Project/Laser-Harp/690 $\endgroup$ – Chris Aug 13 '12 at 14:29
  • $\begingroup$ Well i thought that tracking the same red laser beam's height was easier than tracking many hands from many users (With Kinect is not that complicated.). Now I'm considering using a infra-red range sensor like this one: infusionsystems.com/catalog/product_info.php/products_id/136 Mainly because i don't know if there will have the same "air quality" for a long time so that the beam will be visible and consistent solid. I still want to test the laser solution thought as the visual feedback for the user and for the machine is the same thing. $\endgroup$ – Ventolinmono Aug 13 '12 at 17:49
  • $\begingroup$ That Make project looks very cool. Too much electronics though. It will make a nice prototype but a solution with less parts can be more robust for an end-user release. $\endgroup$ – Ventolinmono Aug 13 '12 at 17:59

I can only say something about the computer vision part.

If your doing color tracking you have to take different lighting conditions into consideration, although it will be easier with a laser (compared to other things) since it at least have the same color.

If you are to use this device professionally you should have a delay about 20ms else it will be noticeable. This forces you to use a high speed camera of at least 100fps (10ms per frame) to have time for processing and sending the sound through the amplifiers.

Since you are only tracking straight lines I would recommend to first filter out everything but the laser-color and then hough transform it to find all straight laser lines. And as a final step implement an ad-hoc method to find where the laser ends shouldn't be that hard.

  • $\begingroup$ Yes i will be having some control of the lightning conditions. And the red color of the laser can be relatively easy to track. The camera speed is an issue i haven't considered, thanks. $\endgroup$ – Ventolinmono Aug 9 '12 at 21:29
  • $\begingroup$ Also i have to be very careful to make the laser beam visible with a fog machine. ¿Any other methods to do this? $\endgroup$ – Ventolinmono Aug 13 '12 at 13:30

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.