Forum

Author Topic: Camera calibration  (Read 11673 times)

matko

  • Newbie
  • *
  • Posts: 8
    • View Profile
Camera calibration
« on: November 12, 2019, 03:27:04 PM »
Hi Everyone, I sent this message before but I think it has been deleted.

I did a 1km per 500m survey in a very flat area with a DJI-Phantom 3 Pro.

The problem is that when processing the information with Agisoft Metashape the results are somewhat bad.

Altimetrically it deforms the model like a parabola, with differences of up to 10m in heights, when in fact on the ground the altimetric differences do not exceed 1m.

I was reading in this and other forums, I also read publications about it and everything indicates that the problem is due to the lack of camera calibration.

I tried unsuccessfully to calibrate the lens (within the Agisoft software) and calibrate the camera, the results change, but they do not improve substantially.

They could help me with this issue?

From already thank you very much.
Best regards

Alexey Pasumansky

  • Agisoft Technical Support
  • Hero Member
  • *****
  • Posts: 14813
    • View Profile
Re: Camera calibration
« Reply #1 on: November 12, 2019, 04:52:46 PM »
Best regards,
Alexey Pasumansky,
Agisoft LLC

matko

  • Newbie
  • *
  • Posts: 8
    • View Profile
Re: Camera calibration
« Reply #2 on: November 14, 2019, 04:31:10 PM »
Thanks Alexey, I couldn't see where the previous messages were, I will detail you:

- The first processing I did without GCP, to visualize the data, you can see that it deforms the terrain but it seems normal to me since there is no control.

- The second processing I did filtering some photos that were not of good quality and worsened the final results.

- The third processing I did with GCP and you can see that the results improve somewhat but not enough.

After this I began to investigate the details of the lens and the camera and I made new tests, in: tools \ lens \ calibrate lens ...
and then in tools \ camera calibration ... I can see that it recognizes the characteristics of the camera (FC300X [3.61mm]).

- In the tests I did using:
«Camera Calibration / Rolling-Shutter Compensation» [on]
«Camera / GPS / INS offset calibration» Enable Reference [on]
«Camera / GPS / INS offset calibration» Adjust GPS / INS offset [on]

The results change but not for the better ...
I also tried «Tools» / Optimize Cameras ...

I was reading about these types of problems and what I could find are articles like this:
https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2011JF002289
That suggest changes for improvements but since the Survey, which I can't do now. I would like to know if I can take advantage of my data

Спасибо большое!

Shipment attached screenshots:

matko

  • Newbie
  • *
  • Posts: 8
    • View Profile
Re: Camera calibration
« Reply #3 on: November 19, 2019, 07:33:13 PM »
How about Friends: unfortunately I have not been able to resolve this issue since processing, apparently there is no camera model that allows me to correct the curvature effect.
I will opt for a polynomial interpolator that allows me to correct the already generated dsm.
Thank you for your contributions.
Regards

Alexey Pasumansky

  • Agisoft Technical Support
  • Hero Member
  • *****
  • Posts: 14813
    • View Profile
Re: Camera calibration
« Reply #4 on: November 19, 2019, 10:14:10 PM »
Hello matko,

Are you able to share the project in PSZ format with the alignment results with support@agisoft.com?
Best regards,
Alexey Pasumansky,
Agisoft LLC