Camera Calibration Method & Parameters

[Geo]

Greetings,  I found the following comment regarding camera calibration that questions the accuracy and limited paramaters used in the Browns model:

"I should also say that Agisoft has fewer parameters in its lens model than does ADAMTech. For many applications this is fine, but for high-accuracy geomatics work it's crucial, in my opinion, to have all lens parameters activated and long floating point representations of the parameters (many programs appear to truncate the values)."

Is this true that the Browns model does not accurately model all of the needed lens distortion parameters if the appropriate floating point representations are used? 

[Porly]

Hello,

Which model ADAMTech is using???

By the way, it would be very useful (in PhotoScan or Agisoft LENS) to run the calibration procedure with auto-targets. The result would be much better, since automatic ellipse operator (recognition) works more accurate (1/50 px) than the matching of (natural-)features in the alignment step. Calibration step in Agisoft LENS is for very low range and on a flat surface (screen). With calibration by auto targets we could create our own testfield in a simple way and add some points with different depth information in object direction.

Only a idea....

Best regards

[kiteman]

very interesting! Could you explain better how to make this your idea?