Is there a link between the intrinsic parameters and the Z16 calculation?

[MGeli]

Hi,

After performing the calibration of a D410 device with Intel's CalibrationTool, we noticed the intrinsic parameters returned via rs-enumerate-devices changed slightly and the accuracy of the Z16 values in the depth frames improved.

I wonder if there's a link between the changes in the intrinsic parameters and the improved accuracy of the Z16 values.

In other words, do the intrinsic parameters affect the calculation of Z16 values?

If I were able to "manually" change the intrinsic parameters inside the device with "random" values, would the device return bad Z16 values?

Thank you.

[Alexandru_O_Intel]

Hello MGeli, Thank you for your interest in the Intel RealSense D410 module. The described behavior is expected when performing a camera calibration. You can find more information on the intrinsic parameter in the link below: https://github.com/IntelRealSense/librealsense/wiki/Projection-in-RealSense-SDK-2.0#intrinsic-camera-parameters The intrinsic parameters of the module are unique to the module, and changing these parameters to random values will result in a bad calibration table. Please let me know if you require any further information. Best regards, Alexandru

[MGeli]

Hello Alexandru,

Thank you for the information.

Unfortunately, the link provided doesn't mention any relationship between the intrinsic parameters and the Z16 distance calculation.

But you confirm that the intrinsic parameters affect how the Z16 distance is calculated, right?

If I were to slightly change those values, I would get different Z16 values, right?

Thank you,

Marc

[Alexandru_O_Intel · Accepted solution]

Hello MGeli,

The intrinsic parameters are: width, height, ppx, ppy, fx, fy, model and coeffs.

When you are not getting the correct intrinsic values, the whole measurement will be affected.

For example, if you want to use the deprojection operation, if you take the wrong 2D coordinates and the depth value, the outcome will be changed.

RMS Error (or Spatial Noise) measures the intrinsic variation in depth values across an ROI (region of interest) assuming a planar target. You can learn more about it on page number 9, from Depth Quality Tool -

https://www.intel.com/content/dam/support/us/en/documents/emerging-technologies/intel-realsense-technology/RealSense_DepthQualityTesting.pdf

Best regards,

Alexandru

[MGeli]

Thank you Alexandru for the information.

I understand your example, it makes sense.

Thank you,

Marc