Subjective speckle suppression in laser based stereo-photogrammetry

Collapse
X
Collapse
 
  • Page of 1
  • Time
  • Show
Clear All
new posts
Previous template Next
  • HansWellens
    3Dfollower
    • Nov 2016
    • 21
    #1

    Subjective speckle suppression in laser based stereo-photogrammetry

    Subjective speckle suppression in laser-based stereo photogrammetry
    https://www.spiedigitallibrary.org/journals/Optical-Engineering/volume-55/issue-12/121713/Subjective-speckle-suppression-in-laser-based-stereo-photogrammetry/10.1117/1.OE.55.12.121713.short?SSO=1
    The use of objective speckles as patterns is of high interest for the ongoing development of stereo photogrammetry. The depth of focus of the projected speckle patterns, which can be found to be several meters, can hardly be matched by other projection principles. On the downside, the use of coherent light leads to subjective speckles generated by the rough surface of the object under test. This effect decreases the accuracy under which objects can be reconstructed. We show how laser-based stereo photogrammetry can be adjusted to increase the measurement accuracy of three-dimensional (3-D)-surface measurements while preserving the advantages of speckles projection. Therefore, we present a method to decrease the contrast of subjective speckles in the images by pixel-wise shifting the cameras orthogonally to their viewing direction and back shifting the taken images numerically, accordingly. This leads to an increase in 3-D-reconstruction quality, as seen in a decrease in standard deviation, peak-to-valley value and in an increase in the number of reconstructed points for measured test objects.


    Its paywalled (not surprisingly), but since this is your area of interest, I'm quite sure you can get it somehow.

    My question is whether someone of your crew of wizards could tell me whether there is any merit in the techniques explained.
    If so, if there would be any chance of this ever ending up as a feature in Zephyr Pro? I can't be the only one using dot pattern illumination, can I?

    With kind regards,
    and keep up the good work!

    Hans
    Zephyr Pro user.
    The left photo contains the random pattern laser illuminated immage. The "pitting" due to the high contrast laser dots can readily be observed.
    1 Photo
    Tags: None
    • Andrea Alessi
      3Dflow Staff
      • Oct 2013
      • 1335
      #2
      Hi Hans!

      well, since your rig is fixed, have you tried using the XMP workflow?

      This way, you have to do camera orientation just once, as they will not move between one capture and the next. So you do not need to project the pattern when actually scanning the users.

      Tutorial here: https://www.3dflow.net/technology/do...flow-tutorial/

      As for the paper, that's more Roberto's cup of tea, so i'll forward this post to him

        Comment

        • HansWellens
          3Dfollower
          • Nov 2016
          • 21
          #3
          Hi Andrea,

          I did test the XMP workflow, and it seems to work very well! A few points of interest: for the XMP workflow to work, the photo files should not include sequential numbering; just the camera numbers (e.g. CAM01, CAM02, ...). It seems to me that it did not work when taking photos in batches (like CAM01_011, CAM02_011, CAM03_011, ...). That could be something for a future update (in which case a predetermined file naming format should be put forward).
            👍 1

            Comment

            • Andrea Alessi
              3Dflow Staff
              • Oct 2013
              • 1335
              #4
              Hi Hans,

              glad to hear the xmp workflow worked well!

              I'm not aware of an issue with that type of suffix (in fact, the only thing we have in place for suffixes is that _mask/_masked will be treated as masks and _something and _somethingelse will be treated as two different layers for multilayer datasets).

              May ask you some details so i may try to replicate the behaviour? thanks!

                Comment

                • cam3d
                  3Dflover
                  • Sep 2017
                  • 682
                  #5
                  HansWellens Hey! I've had some experience building multi camera rigs - Have you explored using a projector as an alternative to laser points? - With the laser points I suspect with each frame, you're blowing out the detail at each point on the surface (causing the pitting effect).

                  I'd recommend a very bright projector so you can keep your shutter speeds fast - Just project a random noise pattern (or granite texture) and keep the projected image as sharp as possible. - If you ever wanted to capture the colour data for your scans you can take two sets of photos in quick succession (under 1sec - as fast as possible), one set with projection on the surface, the other illuminated with diffused strobes- Then you can utilise the XMP workflow to great effect. Best of both worlds: mesh with projected surface and texture with strobe illumination.

                  I've had great success using projectors in the past! But always bought cheaper, less bright models, & regretted it.





                    Comment

                    Previous template Next
                    Working...