Forum

Author Topic: Aerial Photos of Offshore Waters Won't Align  (Read 6424 times)

samblar1

  • Newbie
  • *
  • Posts: 1
    • View Profile
Aerial Photos of Offshore Waters Won't Align
« on: May 02, 2017, 06:17:14 AM »
I'm currently working on a project where I have to stitch together large amounts of photos taken with a drone of different areas right off shore. When I run the program I always narrow down the photos to about 80 or so, so I don't run into any problems with 500+, However, most are still coming out like different variations of the photos attached. I thought that maybe it was due to some glare on the water in the pictures, but when I cropped the glare out of the pictures that only helped with a small portion of the photos. Now I'm thinking it might also have something to do with the ripples in the water not being able to align properly or something, but I'm not sure I just know that the probably seems to be the photo alignment step. Has anyone else run into any problems like this and have any suggestions?

SAV

  • Hero Member
  • *****
  • Posts: 710
    • View Profile
Re: Aerial Photos of Offshore Waters Won't Align
« Reply #1 on: May 02, 2017, 07:38:48 AM »
Hi Samblar1,

Agisoft Photoscan and all other digital photogrammetry software packages cannot deal with non-stationary scenes/objects (such as the sea, which has constant wave activity and is moving). The Structure-from-Motion (SfM) algorithms that are employed cannot handle the changes in the overlapping images, hence the errors/spikes you see in your model.

Glare (as you mentioned), reflective and untextured surfaces are all problems for SfM algorithms.

Sorry for the bad news, but you won't be able to generate an accurate surface for large bodies of (moving/non-stationary) water.

Regards
SAV

Alexey Pasumansky

  • Agisoft Technical Support
  • Hero Member
  • *****
  • Posts: 15746
    • View Profile
Re: Aerial Photos of Offshore Waters Won't Align
« Reply #2 on: May 02, 2017, 07:49:38 AM »
Hello samblar1,

Can you provide the list of parameters used for the processing and also the screenshot of the model view with the camera positions and bounding box orientation?

If the screenshot represent the orthomosaic, it seems that it has been generated based on height field mesh reconstructed with the incorrect bounding box orientation.
Best regards,
Alexey Pasumansky,
Agisoft LLC