4 Replies Latest reply on Nov 4, 2017 11:51 PM by geoffmciver

    Improve depth sensing for IR absorbent material

    KFindlater

      I am finding that certain materials (like bread..) are absorbing a lot of IR light and are missing a lot of depth data whereas other materials/objects even like my hand give nice full depth return (no big black holes) on the Intel Euclid's depth camera and also the SR300.

       

      Would an ambient IR floodlight (not too bright, but still reasonable) help provide enough return light for the stero vision to provide depth data for these materials?

        • 1. Re: Improve depth sensing for IR absorbent material
          MartyG

          I admit that it is the first time that I have heard of bread being resistant to IR scans!  I could not find a lot of information on this topic, but evidently porosity of a material can affect laser scans because a 178 page science paper was written about it.

           

          http://www.measurement.sk/2016/Marynowicz.pdf

           

          Skipping ahead to the conclusions on page 178, they did not offer a solution for the problem though.

           

          Given the influence of porosity on a scan, I wonder if the easiest approach would be the old trick of covering the object with a spray-on powder like foot powder or baby powder to give the camera more to lock on to.

          • 2. Re: Improve depth sensing for IR absorbent material
            KFindlater

            Marty,

             

            Not only is the material porous, it's generally not very 'shiny' (low reflectivity). if the bread was wet, or oily, it would help, but only for specular reflections and even those may end up causing over-exposure in those spots.

             

            Even line laser scanner (triangulation method) have signal return issues if the laser isn't powerful enough to get enough return from bread.

             

            Anyway the realsense technology uses laser projects with structured light patterns in addition to ambient light, and depends on the ASIC and algorithm involved which may be different for each camera as well. So I'm wondering if for the provided examples of the SR300 and the Euclid would a diffused IR flood/ring light help or hinder the operation? 

             

            I figure I can set the realsense devices to use outdoor mode to force them to rely on ambient IR light rather than their little underpowered laser projectors. I figure there'd be a large different in quality/ results from planar or non-textured surfaces in this situation though, where the projected structured light helps the most

            • 3. Re: Improve depth sensing for IR absorbent material
              MartyG

              I remember a case where a user had an 'LED wall' that they said provided uniform white light with no shadows.

              • 4. Re: Improve depth sensing for IR absorbent material
                geoffmciver

                If its absorbing the IR from the projector of the SR300 I'd try a ZR300 with an external IR light.

                 

                The ZR300 works well outdoors using IR from the sun.  The sun easily overpowers the laser IR projector on the ZR300.

                 

                Remember that the ZR300 is effectively a stereo camera. Indoors it uses a IR projector to illuminate and give texture to objects to better determine their disparity in the IR cameras.