Camera parameters for a 152.4mm camera with scanned 9 inch film

[Nicolas_Cadieux]

Hi,

I am refining a work flow for historical scanned images. Theses are from the 1950’s. We are talking about aerial photos taken a 20000 feet with a 152.4mm camera on 9 inch film. The scans are 10680 x 10680 pixels at 1200dpi. The ground sampling is about 80cm.

The current “best guess” by ODM for the camera.json file is the following.

{
" 10680 10680 brown 0.0": {
“projection_type”: “brown”,
“width”: 10680,
“height”: 10680,
“focal_x”: 0.7568916569184332,
“focal_y”: 0.7568916569184332,
“c_x”: 0.008747737758061574,
“c_y”: 0.015261935056106802,
“k1”: 0.005902635597588758,
“k2”: 0.0054292388794959165,
“p1”: 0.0001354910146991932,
“p2”: 0.0023083494658548793,
“k3”: -0.013422795667489532
}
}

Is there a way I can improve on this best guess camera model? Where can I find precise information on these variables?

Thanks
Nicolas

[Saijin_Naib]

OpenCV has some documentation on Camera Distortion modeling here: OpenCV: Camera Calibration

This is some serious math, though.

From what I’ve been reading, these coefficients are standarized for Camera Distortion modeling and not specific to OpenCV (and OpenSFM/OpenDroneMap by extension), so other resources should also suit.

The real question is, given the age of the camera platform, were these values ever measured or published so that you can use them to improve the reconstructed Camera Distortion model?