Suitable surface for keypoints

I have taken a number of photos of a 400m long bridge with a UAV and have had problems getting it to calibrate. On inspection it seems that most of the keypoints for the images are coming from the surrounding vegetation with hardly any coming from the bridge itself (see images). The bridge is wrought iron with an iron deck with gravel infill (see close up). Is it that the photo was taken to far away…too close or is it just that this is not a suitable subject for photogrammetric modelling?  .

Many thanks in advance for any help.

Paul

Hi Paul, 

There are many parameters to take into account to understand why there were issues during calibration. Was it possible to get a Point Cloud reconstruction of the bridge? or did the processing stop before? Next to the image with the orange crosses, there is a slider that shows how many images were used for each orange cross, what was the value of this slider in your screenshot? If it was high, it means that there could be other matches displayed, if it was low (e.g 2) then there were no matches found on the rails.

Maybe the repetitive pattern of the rails and bridge could make the calibration process more difficult. 

More files that could help explain the calibration problem, would be:

  • Quality Report of the project (click Process > Generate Quality Report)
  • Log file (Text file in the results folder)

Cheers, 
Pierangelo