On Criteria for Wide-Angle Lens Distortion Correction for Photogrammetric Applications

Abstract

Photogrammetry is intensively used both in industrial applications and medicine. Extraction of various anthropometrical data, such as height, shoulder inclination or Kobb angle form single photograph could be implemented in mobile devices, used for health screening. To be able to get photo of standing individual when the space for observations is restricted, wide - angle lens has to be used. Unfortunately, such lenses demonstrate highly expressed radial distortion, thus image has to be corrected before it could be used for estimation the subject’s dimensions. There are number of methods, proposed to correct lens radial distortion. Most of them use deviation of the imaged straight line from the real straight line as a criterion for image correction. Correction parameters are adjusted to minimize the deviation. When one have to select correction method for particular lens and particular application, it could be reasonable to choose method that provides minimal, comparing to others, residual deviation. Besides, this paper demonstrates, that such criterion does not guarantee that error of dimensions measurements will be minimal as well. In the present work, various radial distortion correction methods, including one, introduced by authors, were evaluated, using both straight line deviation and dimension measurement error. Obtained results allowed formulate recommendation for improvement of existing image correction methodology.