A Method for Lens Distortion Correction of Algorithmically Altered Images

Photogrammetry is a commonly used type of analysis in accident reconstruction because it allows the position and orientation of vehicles to be quantified as seen in photographic and video evidence. Lens distortion is an important consideration when analyzing photographs and video through photogrammetry. Failure to account for lens distortion can result in inaccurate spatial measurements, particularly when elements of interest are located toward the edges and corners of images. A variety of methods for removing lens distortion are commonly used in photogrammetric analysis. However, these methods assume that lens distortion is solely the result of curved camera lenses and has not been altered algorithmically by the camera system. Today there are several cameras on the market that algorithmically alter images before saving them. These camera systems use proprietary distortion correction features to visually change photographs and videos before they can be viewed. A video or photograph produced by these types of camera systems can display unexpected distortion patterns which cannot be corrected using common methods. This paper presents a methodology for correcting the type of non-uniform lens distortion seen in this type of video. Using the Rove R2-4K Dash Cam, which utilizes a Proprietary Distortion Correction Algorithm, as an example, we explain why common lens correction methods cannot be used for videos of this nature and evaluate an alternative method that allows for consideration of unexpected distortion patterns resulting from the algorithmic correction. The method presented can be utilized with image or video files containing any type of distortion, even if it has been altered by the camera system.