Matlab code for Super-Wide-Angle-lens Radial Distortion correction just using a single image of a checkerboard. The implemented procedure is fully automatic, without any necessity of manual selection of corners or parameters.
Xianghua Ying, Xiang Mei, Sen Yang, Ganwen Wang, Jiangpeng Rong, Hongbin Zha
Peking University, Beijing 100871, P.R. China
Email: xhying@cis.pku.edu.cn
Version Matlab Source Code Image Data
2.1 SWARD_Toolbox_2.2.zip Image.zip
Notice: In order to run the toolbox, you may install Matlab version R2010a or later (tested under Matlab 2012a).
Copyright: This software is made publicly for research use only. For commercial applications please contact xhying@cis.pku.edu.cn to discuss alternatives.
Simple user's guide: PDF

Citations

Flag Counter

Description

In Hartley-Kang's paper, they proposed a very efficient method to determine the center of radial distortion by finding the epipole in a radial distorted image of a planar calibration pattern, whereas the original planar calibration pattern is directly treated as an image to construct an image pair together with the radial distorted image. This paper presents a convex optimization method for radial lens distortion correction from a single image of a planar pattern using the monotonicity constraints of radial distortion function. In the Hartley-Kang's method, it is required about twenty images of a planar pattern taken in different orientations and positions, to obtain satisfactory results of radial distortion correction using a least square method. However, in this paper, we illustrate that only utilizing a single radial distorted image of a planar pattern, reasonable corrected results can still be obtained by using our method, and the results may be better than those from the Hartley-Kang's method in the same situation. More information please refer to the citations.

Index Terms - Radial distortion, Distortion correction, Convex optimization, Camera calibration, Fisheye lenses, Checkerboard, Single image, video surveillance

References