Determination of the Center of Radial Distortion for a Camera Lens

Pablo Vera, Octavio Icasio-Hernández, Joaquín Salas

Research output: Contribution to journalArticlepeer-review

5 Scopus citations

Abstract

A precise determination of the center of radial distortion (CoD) improves the characterization of the distortion introduced by the lens, and hence, supports a better description of the perspective projection process. Nonetheless, its value is often neglected and approximated by the center of the image. In our research, we aim to determine a lens CoD with high accuracy using derivative instruments. To that end, we introduce a method that bounds the region containing the CoD within one pixel of uncertainty by characterizing the residual of a straight line (RSL) fitting process. The procedure is further refined to obtain an estimate for the CoD position by first inferring straight lines (ISL) crossing it and then finding the intersection between these lines. Later on, we explore the introduction of the division model assumption (DMA) for lens distortion in the estimation of the CoD. The RSL approach becomes relevant as we compare the ISL and DMA methods with alternatives from the literature as it provides a reference value containing the CoD. In our experiments, we take images of a straightness gauge by moving a camera with a coordinate measuring machine at regular and accurate steps. At each location, we estimate undistorted lines at different orientations of the camera and find the CoD from their intersections. Our approach permits the establishment of the CoD with high accuracy for a camera lens with radial distortion.

Original languageEnglish
Article number8999531
Pages (from-to)7015-7027
Number of pages13
JournalIEEE Transactions on Instrumentation and Measurement
Volume69
Issue number9
DOIs
StatePublished - Sep 2020

Keywords

  • Center of distortion
  • coordinate measuring machine (CMM)
  • lens distortion
  • straightness gauge

Fingerprint

Dive into the research topics of 'Determination of the Center of Radial Distortion for a Camera Lens'. Together they form a unique fingerprint.

Cite this