Forum

Author Topic: Align images ONLY using GCPs acquired by total station  (Read 5187 times)

speedbrake

  • Newbie
  • *
  • Posts: 16
    • View Profile
Align images ONLY using GCPs acquired by total station
« on: June 26, 2013, 03:19:36 PM »
Hi
we are working in a university project aimed to estimate some statistics on errors comparison between a Photoscan Pro and a laser scanner survey of the same building.
We have collected more than 160 images captured from ground and by UAV including EXIF data with GPS tag (standard gps receiver).
We have made a very dense laser survey and, using a total station (teodolite) , we have also collected more than 100 GCPs around the building and ground.
The 100 GCP are all referred to one Cartesian referring system (one of the 4 different positions of total station during acquisitions) and the laser data are now referred to this system and all is ready to be compared to the Photoscan Pro model
Now we are looking for the best way to align images and georeferring these.
The tutorial you published, uses camera positions stored in a file and GCPs to optimize the model.
In our project, we don’t have a similar file and we have only poor information of camera positions (gps).
We could image to follow two ways:
1)Align images without EXIF import
2) Build geometry as cloud point
3)Detect 10-15 GCPs on images
4)Import the GCPs coordinates list
5)Optimize (for camera position and point cloud reprojection)
6) Make some statistics measuring rms with the remaining GCPs unused
but in this case the referring system, adopted by PS during align, is different from the coordinate system I import
The second way could be
3) ,  4) , 1)
so the imported system and that used from PS to align should be ugual….but I didn’t find any evidence about this procedure inside you user guide.
Could you please indicate us the correct way to obtain the best referred model related on very precise GCPs acquired by the total station?
Thanks
paolo

Alexey Pasumansky

  • Agisoft Technical Support
  • Hero Member
  • *****
  • Posts: 14813
    • View Profile
Re: Align images ONLY using GCPs acquired by total station
« Reply #1 on: June 26, 2013, 03:45:00 PM »
Hello Paolo,

The workflow similar to your tasks is described in the following Tutorial: http://www.agisoft.ru/tutorials/photoscan/06/

1) Align photos using ground control pre-selection based on GPS camera locations (coordinate system is set to WGS84)
2) Build rough geometry,
3) Place markers,
4) Import markers coordinates and switch coordinate system to Local coordinates (or coordinate system that corresponds to the GCPs). do not forget to uncheck all cameras in Ground Control pane,
5) Press Update button and/or optimize the camera alignment,
6) Perform other tasks (Build geometry, export, etc.)

In case you do not have any coordinates for cameras you can use Generic pre-selection on the first step. Using pre-selection is recommended since it reduces the processing time significantly for large datasets.
Best regards,
Alexey Pasumansky,
Agisoft LLC