Forum

Author Topic: Camera calibration for UAV  (Read 5070 times)

photoscan_user

  • Jr. Member
  • **
  • Posts: 98
    • View Profile
Camera calibration for UAV
« on: May 29, 2018, 12:24:26 PM »
I have a camera that used in UAV flights and I want to calibrate it.

In flight camera focus is set towards infinity, so when I try to calibrate camera I use this setting, but in this setting chessboard looks blured when I move camera in front of screen to cover full screen frame, calibration process seems works fine but do I do everything right?

Also can you clarify in Camera Calibration settings what is difference between Focal length (mm) and f parameter (according to documentation it's in pixel and also called focal length), is the same focal length just in different units or it's mean something different?

mm1992

  • Newbie
  • *
  • Posts: 5
    • View Profile
Re: Camera calibration for UAV
« Reply #1 on: May 29, 2018, 05:52:22 PM »
I calibrated my UAV by putting the chessboard on a large TV (70") and standing back a few feet until the photos were in focus.  It didn't have full coverage, but I made sure to get photos with it in each corner so I end up with effectly gettting full coverage.  The results seemed to work just fine from that.

photoscan_user

  • Jr. Member
  • **
  • Posts: 98
    • View Profile
Re: Camera calibration for UAV
« Reply #2 on: May 30, 2018, 12:34:51 PM »
Also if I have 2 cameras of same model is it reasonable to use different calibration for each of them or it's ok to use averaged calibration?

SAV

  • Hero Member
  • *****
  • Posts: 709
    • View Profile
Re: Camera calibration for UAV
« Reply #3 on: May 31, 2018, 05:55:00 AM »
Hi photoscan_user,

Yes, f stands for focal length, but in the camera calibration window it is mentioned in two different units.
mm (at the top)
pixel (further down)

You can easily calculate the f (mm) value by multiplying the pixel size mentioned in the camera calibration window with f (pixel).

Regarding camera calibration. If you are using accurate reference information such as GCPs that were surveyed at high accuracy (RTK/PPK GPS), then you don't need to calibrate your camera beforehand but instead let PhotoScan perform a self-calibration during processing.

All the best.

Regards,
SAV



Alexey Pasumansky

  • Agisoft Technical Support
  • Hero Member
  • *****
  • Posts: 13289
    • View Profile
Re: Camera calibration for UAV
« Reply #4 on: May 31, 2018, 11:52:13 AM »
Hello photoscan_user,

Usually pre-calibrated values are reasonable only for high quality metric cameras, whereas for compact low grade consumer cameras it may lead to even worse results. For example, calibration for cameras with slow rolling shutter will be practically useless, as the rolling shutter distortions would depend on the camera movement and rotation speed.

So in order to rely on the precalibrated results it is necessary to perform at least few validation experiments.
Best regards,
Alexey Pasumansky,
Agisoft LLC

Yoann Courtois

  • Sr. Member
  • ****
  • Posts: 307
  • Engineer in Geodesy, Cartography and Surveying
    • View Profile
Re: Camera calibration for UAV
« Reply #5 on: June 05, 2018, 11:26:14 AM »
Hi !

It is well incoherent to pre-calibrate a UAV camera using a chessboard on a screen, even on a large TV :)

As photoscan_user said, best consistent results are obtained while focusing toward infinity.
Consequently, even if sharp images may be observed (by our eyes) when taken 5 meters far from the screen, a one-pixel-sharpness cannot be obtained closer than 10-12 meters, or even 15 m in a gloomy room
(Results coming from calculation using Phantom 4 Pro features)

So either you got a 15m x 22m screen, or this process is impossible.

But furthermore, pre-calibrate such commercial cameras with lots of small distortions (fitted by k3-k4 & p3-p4) wouldn't be accurate with a 15x22 chessboard where only square corners can be used. It would indeed required big enough squares to be detected, and so not a really huge amount of points.

Conversely, a large and well-overlapped survey with an overabundance of accurate GCPs would allow you to calculate in-situ calibration that will fit the whole sensor, together with a thousands redundancy !

Nevertheless, keep in mind that such sensor may suffer changes depending on survey weather conditions. Then, a well-determined in-situ pre-calibration will help you a lot to keep a small survey a good stability, but above a certain number of pictures, release inner parameters will be better to fit the most to the daily camera alterations !

Regards
--
Yoann COURTOIS
R&D Engineer in photogrammetric process and mobile application
Lyon, FRANCE
--

ruyi7952

  • Full Member
  • ***
  • Posts: 103
    • View Profile
    • My Blog
Re: Camera calibration for UAV
« Reply #6 on: June 05, 2018, 06:40:58 PM »
Agree with SAV / Alexey Pasumansky

I suggest that the camera calibration parameters can be obtained by means of self-calibration in the way of ground control points.

PhotoScan has built-in coding target printing function to output appropriate control point target.

My current working experience is that I completely rely on self-calibration of PhotoScan and do not deliberately set ground control points for camera calibration, thus obtaining 3d model data with accuracy below ±5cm.

If I do it deliberately, I am confident to do it plus or minus ±1 cm.
Real Scene Model Professional Contractor

Collect、Process、Publish

Bill68

  • Newbie
  • *
  • Posts: 4
    • View Profile
Re: Camera calibration for UAV
« Reply #7 on: April 08, 2019, 09:20:43 PM »
Is it possible to over optimize the camera calibration?

I flew a small test project with a Phantom 4 RTK with about 100 photos at 400 foot elevation.  There are 6 high accuracy GCP's.  My original plan was to use the GCP's only as check points, but residuals were a foot off.  I decided to optimize the camera calibration and now the GCP's check within 0.01 foot, and camera coordinates show about 0.02 foot, which seem too good to be true.

Is there a danger that the camera calibration simply fit itself to the data?