Since we’re often hired long after the collision, accurate photogrammetry analyses are crucial for private recons. When direct access to the vehicles or scene isn’t possible, available photographs are often used to model crush and roadway evidence.

Every lens introduces distortion to some degree, so the more accurately we handle that distortion, the better the photogrammetry analysis. Ideally, we would track down the camera used to generate the images of interest and perform a calibration - a process that maps the distortion. To achieve this, a calibration sheet (see image below) is photographed from varying angles and a computational routine calculates the distortion as a function of the distance from the principal point (i.e., the effective center).

 
 

Since the original camera is rarely available, we have to rely on other ways to correct distortion. In my experience, one of the best methods is to calibrate an exemplar camera and lens (if detachable). Adjust the settings on the exemplar camera to match those of the subject camera, as closely as possible, capture 12+ photos of a calibration sheet, and feed those into a calibration algorithm, like that available in PhotoModeler. When the subject photographs are brought into a new photogrammetry project, the exemplar calibration can be applied and accuracy should be significantly improved.

This is a method I’ve been using for over a decade with great success, and I know of many colleagues who do the same. That said, I’m not aware of any research examining how distortion varies between cameras of the same make and model. I recently calibrated 16 Nikon Coolpix S33s to better understand distortion variability and expect the results to be published later this year. I’ll share the results once everything is buttoned up.

If you’d like to learn more about performing a sheet calibration, this video from PhotoModeler shows how to take the photos and this one shows how to process them. For those wanting a deeper dive, I’ll be offering a live three-day photogrammetry course this May in Tempe and will share registration details shortly.

Thanks for reading, keep exploring!

Lou Peck
​Lightpoint | Axiom