I have a problem ragarding transferring a point cloud to a 2.5D metric grayscale image (gray value gives depth information). I start with a three channel image of 32bit floats with each pixel representing the three coordinates in mm (metric).

Halcon is the preferred library, but not necessary.

I hope that someone can help me.

Greetings from Stuttgart, Germany


  • $\begingroup$ What was the problem? $\endgroup$ – D.J.Duff Apr 3 '15 at 5:27
  • $\begingroup$ I do not know how to transfer a point cloud to 2.5D metric grayscale image. (operators??) $\endgroup$ – e_mka Apr 7 '15 at 14:08
  • $\begingroup$ Here's a good place to start: en.wikipedia.org/wiki/3D_projection $\endgroup$ – D.J.Duff Apr 8 '15 at 5:03
  • $\begingroup$ If I misunderstood the question, I'm sorry. $\endgroup$ – D.J.Duff Apr 8 '15 at 5:03

Your 3 channels already give you the 3D coordinates, and from that 3 images, you already know which pixels locations the 3D data are projected. So you don't need to do anything. If you like to get a depth image, just set: $$ depth(i,j) = z(i,j) $$ In Halcon, you just select the Z-channel using access_channel. If you desire a distance image, then: $$ distance(i,j) = \lVert (x,y,z)^T \rVert $$ The norm defines the distance metric you would like to use. Traditionally, I would say Euclidean $\|\cdot\|=\sqrt{x^2+y^2+z^2}$. You could do this in any language by using operations like addition, multiplication and square roots.


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.