Forum

Author Topic: Underwater mapping model curved.. help?  (Read 1980 times)

whowell

  • Newbie
  • *
  • Posts: 2
    • View Profile
Underwater mapping model curved.. help?
« on: July 03, 2019, 01:01:30 AM »
We recently completed two coral reef mapping missions utilizing an autonomous underwater vehicle with two goPro hero7's attached 18" apart.  The vehicle kept a depth of 3-4 meters above the surface, and the water was clear enough to snap decent photos.

There are upwards of 20,000 photos total for the first reef, and after some trial and error I've managed to make two 10,000 photo halves in separate chunks.  Each half consists of 5,000 corresponding photos from each goPro.

As you'll see from the photos, the map displays a prominent curve even though the reefs we were mapping were large flat areas.  I imagine the curve is a product of the extremely long mapping lanes and the goPro fisheye lenses compounded by the refraction underwater.

Is there a way to rectify this?  Would processing the photos in smaller chunks or a different order (not sequentially, but rather the photos that correspond in each of the lanes) alleviate the curve?

xabierr

  • Jr. Member
  • **
  • Posts: 84
    • View Profile
Re: Underwater mapping model curved.. help?
« Reply #1 on: July 03, 2019, 09:08:26 AM »
the use of control points/scale bars will remove the bowl effect.

Alexey Pasumansky

  • Agisoft Technical Support
  • Hero Member
  • *****
  • Posts: 14853
    • View Profile
Re: Underwater mapping model curved.. help?
« Reply #2 on: July 03, 2019, 12:07:31 PM »
Hello whowell,

Can you post the screenshots of Initial and Adjusted tabs from the Camera Calibration dialog?

The bowl-effect could be fixed with some manual input, if you modify the adjusted parameters (mainly the focal length) and optimize  the alignment (without F optimization), but it could require a few iterations.
Best regards,
Alexey Pasumansky,
Agisoft LLC

whowell

  • Newbie
  • *
  • Posts: 2
    • View Profile
Re: Underwater mapping model curved.. help?
« Reply #3 on: July 03, 2019, 04:06:15 PM »
Unfortunately we're not able to place anything on the seafloor, although there are markers placed by the officials overseeing the project for monitoring coral growth.

Alexey, I'm attaching the the screenshot of the camera calibration screen.  What values would you suggest to start with in focal length?  Also, what parameters should we check/uncheck in the Optimize camera action window itself?  I've been unchecking the fit f box, but all the other fit boxes remain checked (besides Fit p3, and Fit p4).  Also, under the advanced tab there are two unchecked boxes Adaptive Camera Model Fitting and Estimate Tie Point Covariance.

I've tried adjusting the focal length from the original 3mm to 5mm, 16mm, 35mm, 85mm successively in the camera calibration screen... and once I Optimize the cameras there is no change to the bowl alignment.

Do I have to re-align the cameras with the new settings or just optimize the cameras?
« Last Edit: July 03, 2019, 06:04:44 PM by whowell »