How to get Ground Truth Depth Image?

• MartyG

  • March 11, 2021 09:17
  • Edited

  Hi Encyrine. The ground truth of RealSense depth measurements can be obtained by adding a minus-value to the measurements.  Pages 72 and 73 of the current edition of the data sheet document for the 400 Series cameras list the minus value to add for particular RealSense models.  There are two separate tables for the models with front glass (D415, D435, D435i, D455) and the caseless Depth Module versions.

  https://dev.intelrealsense.com/docs/intel-realsense-d400-series-product-family-datasheet

  Note that if you have a Depth Module model and are measuring from the back of the module instead of from the camera lens then a positive value is subtracted instead.

  0

  Comment actions Permalink
• Encyrine

  • March 12, 2021 02:40

  

   

  I mean gt-*.png image here. Depth image obtained by camera has many NaN values, unlike gt image here. I want to know the way get clear gt image.

  0

  Comment actions Permalink
• MartyG

  • March 12, 2021 07:49
  • Edited

  Thank you for the additional information.  You could try reducing invalid and nan values by applying a post-processing filter called a Temporal Filter that can decide whether to replace a value with stored historical data.

  https://github.com/IntelRealSense/librealsense/blob/master/doc/post-processing-filters.md#temporal-filter

  The Hole-filling post processing filter can also do pixel corrections, with settings such as having invalid pixels use the value of a neighboring pixel.

  https://github.com/IntelRealSense/librealsense/blob/master/doc/post-processing-filters.md#holes-filling-filter

  0

  Comment actions Permalink

Please sign in to leave a comment.