![](bg8.png)
same radial distortion parameter throughout the tra-
jectory of the robot, as it estimates a new parameter
along with each homography. In order to enforce the
same radial distortion compensation, we may use his-
togram voting of the estimated distortion parameters.
In Figure 6 we have done so for the estimated para-
meters acquired during the parallel parking test se-
quence. The most likely parameter λ
∗
is then used to
rectify a previously unseen image of the calibration
scene. While the rectified image is not perfect, it may
serve as a good initial solution for further non-linear
refinement.
6 CONCLUSIONS
One cannot ignore radial distortion when estimat-
ing trajectories in a Visual Odometry framework. In
this paper we have considered radially distorted ho-
mographies compatible with the general planar mo-
tion model. We have proposed a novel algorithm
for estimating the homographies and showed on both
real and synthetic data that it increases the perfor-
mance compared to a general homography estima-
tion method with radial distortion. Furthermore, we
show, by incorporating the proposed algorithm in a
VO pipeline, that it yields satisfactory results in terms
of estimating the radial distortion parameter.
ACKNOWLEDGMENTS
The author gratefully acknowledges M
˚
arten
Wadenb
¨
ack and Martin Karlsson for providing
the data for the planar motion compatible sequences,
and Magnus Oskarsson for fruitful discussions
regarding the basis selection heuristic which made
the proposed solver faster. This work has been
funded by the Swedish Research Council through
grant no. 2015-05639 ‘Visual SLAM based on Planar
Homographies’.
REFERENCES
Brown, D. C. (1966). Decentering distortion of lenses. Pho-
togrammetric Engineering, 32:444–462.
Chen, T. and Liu, Y.-H. (2006). A robust approach for struc-
ture from planar motion by stereo image sequences.
Machine Vision and Applications (MVA), 17(3):197–
209.
Cox, D. A., Little, J., and O’Shea, D. (2005). Using Al-
gebraic Geometry. Graduate Texts in Mathematics.
Springer New York.
Fitzgibbon, A. W. (2001). Simultaneous linear estimation of
multiple view geometry and lens distortion. In Con-
ference on Computer Vision and Pattern Recognition
(CVPR).
Hajjdiab, H. and Lagani
`
ere, R. (2004). Vision-based multi-
robot simultaneous localization and mapping. In
Canadian Conference on Computer and Robot Vision
(CRV), pages 155–162, London, ON, Canada.
Hartley, R. I. and Zisserman, A. (2004). Multiple View Ge-
ometry in Computer Vision. Cambridge University
Press, Cambridge, England, UK, second edition.
Kayumbi, G. and Cavallaro, A. (2008). Multiview trajec-
tory mapping using homography with lens distortion
correction. EURASIP Journal on Image and Video
Processing, page 145715.
Kukelova, Z., Bujnak, M., and Pajdla, T. (2008a). Auto-
matic generator of minimal problem solvers. Euro-
pean Conference on Computer Vision (ECCV), pages
302–315.
Kukelova, Z., Bujnak, M., and Pajdla, T. (2008b). Polyno-
mial eigenvalue solutions to the 5-pt and 6-pt relative
pose problems. In British Machine Vision Conference
(BMVC).
Kukelova, Z., Heller, J., Bujnak, M., and Pajdla, T. (2015).
Radial distortion homography. In Conference on Com-
puter Vision and Pattern Recognition (CVPR), pages
639–647.
Larsson, V. and
˚
Astr
¨
om, K. (2016). Uncovering symme-
tries in polynomial systems. European Conference on
Computer Vision (ECCV), pages 252–267.
Larsson, V.,
˚
Astr
¨
om, K., and Oskarsson, M. (2017a). Ef-
ficient solvers for minimal problems by syzygy-based
reduction. Computer Vision and Pattern Recognition
(CVPR), pages 2383–2392.
Larsson, V.,
˚
Astr
¨
om, K., and Oskarsson, M. (2017b). Poly-
nomial solvers for saturated ideals. International
Conference on Computer Vision (ICCV), pages 2307–
2316.
Larsson, V., Kukelova, Z., and Zheng, Y. (2018a). Camera
pose estimation with unknown principal point. Com-
puter Vision and Pattern Recognition (CVPR), pages
2984–2992.
Larsson, V., Oskarsson, M.,
˚
Astr
¨
om, K., Wallis, A.,
Kukelova, Z., and Pajdla, T. (2018b). Beyond gr
¨
obner
bases: Basis selection for minimal solvers. Computer
Vision and Pattern Recognition (CVPR), pages 3945–
3954.
Liang, B. and Pears, N. (2002). Visual navigation using
planar homographies. In International Conference
on Robotics and Automation (ICRA), pages 205–210,
Washington, DC, USA.
Ort
´
ın, D. and Montiel, J. M. M. (2001). Indoor robot motion
based on monocular images. Robotica, 19(3):331–
342.
Pritts, J., Kukelova, Z., Larsson, V., and Chum, O. (2018a).
Radially-distorted conjugate translations. In Confer-
ence on Computer Vision and Pattern Recognition
(CVPR).
Pritts, J., Kukelova, Z., Larsson, V., and Chum, O. (2018b).
Radially Distorted Planar Motion Compatible Homographies
287