GeoErnest Response to “Ask me Something” by Mulualem: Do Aerial Camera Self-Calibration, but be Smart

Hi Mulualem,

Should you use self-calibration? Yes! As with any calibration, you have to be extremely careful.

Self-calibration is necessary to model systematic image errors in your aerial camera. This would be focal length, principal point offsets, and the 12 or 44 parameters that describe systematic lens distortion effects.

Self calibration is needed if the camera calibration is incorrect. Once calibrated well, these effects should be well known and should not change much over time (yes, focal length and principal points can change because the metal holding the CCD and the lens on the body can change (over time)… we’re talking microns here… super sensitive.

So how does this work? you’re really using a bunch of parameters to create the most realistic model your camera. And as you know, you can represent this model via a series of equations (I won’t get into this in detail). Like any bundle-adjustment, you need to feed good input data into your model. The hard part is getting the model right, then collecting high-quality data at an accuracy you need to get a good output at the accuracy you want.

Input data could be the known points, whether that be ground control, tie points, points on the wall, etc…

If you look in any textbook, you can see graphics of each of the 12 parameters. They each describe a different effect such as translation, curve left, curve right, bend up, bend down, etc…Together, they form a distortion grid. (I could email you a graphic).

44 parameters is just a more complex way to model the distortion grid (more effects to model). So, yes, 44 parameters is recommended, but that means your calibration environment needs to be more controlled as well. Using 44 parameters is more sensitive to incorrect input information so you have to be more careful. Let’s say you were using flight data to calibrate your camera. If you have a couple of incorrect ground control points and tie points, this is going to cause problems in your calibration (again, bad input, right?) because 44 parameters could absorb any errors into the model.

Back to the original question: Can you trust self-calibration? Yes!

So is terrestrial calibration better than flight self-calibration? Not necessarily. Here’s why:

Terrestrial calibration
– it’s likely indoors in a controlled, static, room temperature environment.
– you are still creating tie points/ground control on the wall (for example)
– However, these are not true flight conditions under vibration, temperature changes (which can change lenses, focus, body) –>not in an operational, aircraft environment
– If you have long focal lengths to calibrate or to check, your depth of field and focus could be a problem if your camera is too close to your wall targets.

Flight calibration
– you are operating in true flight conditions to provide the most realistic calibration environment
– you have to fly a certain flight configuration with high overlaps at different altitudes
– you should fly over areas with many features, such as an urban area
– you need really accurate ground control points and tie points

– The issue is you are no longer in a controlled environment and you have to model all the standard aerial photogrammetry issues like GPS/INS accuracy, quality of ground control points, temperature, vibration, etc.. you have many more things to control and model.

So in summary, yes, use self-calibration because metric camera calibration do change over time (very slowly though). You just have more things in your aircraft environment to model. You have to be careful.

This is GeoErnest.

P.S. As a manufacturer, we do both. We do a terrestrial calibration, then use these results to initialize flight calibration results. The flight calibration is the true test of the entire systems integration.

P.P.S. Any one have comments to add?


About GeoErnest
This is GeoErnest. Visit

One Response to GeoErnest Response to “Ask me Something” by Mulualem: Do Aerial Camera Self-Calibration, but be Smart

  1. Mulualem says:

    Thank You Ernest for your brief answer on the issue. I will like to get the Graphics of the 44 parameters grid, in that way i could picture what sort of parameters are the additional on Ebner’s 12 parameters.
    I was reading a couple of documents about self-calibration and most of them have the 12 parameters grid model but not the 44. so if u can, e-mail me at
    The difference in using and not using the self-calibration can be seen clearly on check point RMS and also more graphically on the camera residual vector graph when u revise Ur statistics in Ur AT software which could implicate the parameters do something in the background when applied.
    Thank You very Much !!!

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: