ITERATIVE DENSE CORRESPONDENCE CORRECTION THROUGH BUNDLE ADJUSTMENT FEEDBACK-BASED ERROR DETECTION

Mauricio Hess-Flores, Mark A. Duchaineau, Michael J. Goldman, Kenneth I. Joy

Abstract

A novel method to detect and correct inaccuracies in a set of unconstrained dense correspondences between two images is presented. Starting with a robust, general-purpose dense correspondence algorithm, an initial pose estimate and dense 3D scene reconstruction are obtained and bundle-adjusted. Reprojection errors are then computed for each correspondence pair, which is used as a metric to distinguish high and low-error correspondences. An affine neighborhood-based coarse-to-fine iterative search algorithm is then applied only on the high-error correspondences to correct their positions. Such an error detection and correction mechanism is novel for unconstrained dense correspondences, for example not obtained through epipolar geometry-based guided matching. Results indicate that correspondences in regions with issues such as occlusions, repetitive patterns and moving objects can be identified and corrected, such that a more accurate set of dense correspondences results from the feedback-based process, as proven by more accurate pose and structure estimates.

References

  1. Duchaineau, M., Cohen, J., and Vaidya, S. (2007). Toward fast computation of dense image correspondence on the GPU. In Proceedings of HPEC 2007, High Performance Embedded Computing, Eleventh Annual Workshop, pages 91-92, Lincoln Laboratory, Massachusetts Institute of Technology.
  2. Hartley, R. I. and Zisserman, A. (2004). Multiple View Geometry in Computer Vision. Cambridge University Press, 2nd edition.
  3. Hirschmüller, H. and Scharstein, D. (2007). Evaluation of cost functions for stereo matching. In IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR 2007), pages 91-92, Minneapolis, MN.
  4. Lourakis, M. and Argyros, A. (2000). The design and implementation of a generic sparse bundle adjustment software package based on the Levenberg-Marquardt algorithm. Technical Report 340, Institute of Computer Science - FORTH, Heraklion, Crete, Greece.
  5. Mayoral, R. and Aurnhammer, M. (2004). Evaluation of correspondence errors for stereo. In 17th International Conference on Pattern Recognition (ICPR'04), volume 4, pages 104-107.
  6. Oxford Visual Geometry Group (2009). Multiview and Oxford Colleges building reconstruction. http://www.robots.ox.ac.uk/ vgg/data/datamview.html.
  7. Rodehorst, V., Heinrichs, M., and Hellwich, O. (2008). Evaluation of relative pose estimation methods for multi-camera setups. In International Archives of Photogrammetry and Remote Sensing (ISPRS 7808), pages 135-140, Beijing, China.
  8. Scharstein, D. and Szeliski, R. (2002). A taxonomy and evaluation of dense two-frame stereo correspondence algorithms. International Journal On Computer Vision, 47(1-3):7-42.
  9. Stewénius, H., Engels, C., and Nistér, D. (2006). Recent developments on direct relative orientation. ISPRS Journal of Photogrammetry and Remote Sensing, 60:284- 294.
  10. Xiong, Y. and Matthies, L. (1997). Error analysis of a realtime stereo system. In IEEE Conference on Computer Vision and Patter Recognition (CVPR), pages 1087- 1093.
Download


Paper Citation


in Harvard Style

Hess-Flores M., A. Duchaineau M., J. Goldman M. and I. Joy K. (2010). ITERATIVE DENSE CORRESPONDENCE CORRECTION THROUGH BUNDLE ADJUSTMENT FEEDBACK-BASED ERROR DETECTION . In Proceedings of the International Conference on Computer Vision Theory and Applications - Volume 1: VISAPP, (VISIGRAPP 2010) ISBN 978-989-674-028-3, pages 400-405. DOI: 10.5220/0002816104000405


in Bibtex Style

@conference{visapp10,
author={Mauricio Hess-Flores and Mark A. Duchaineau and Michael J. Goldman and Kenneth I. Joy},
title={ITERATIVE DENSE CORRESPONDENCE CORRECTION THROUGH BUNDLE ADJUSTMENT FEEDBACK-BASED ERROR DETECTION},
booktitle={Proceedings of the International Conference on Computer Vision Theory and Applications - Volume 1: VISAPP, (VISIGRAPP 2010)},
year={2010},
pages={400-405},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0002816104000405},
isbn={978-989-674-028-3},
}


in EndNote Style

TY - CONF
JO - Proceedings of the International Conference on Computer Vision Theory and Applications - Volume 1: VISAPP, (VISIGRAPP 2010)
TI - ITERATIVE DENSE CORRESPONDENCE CORRECTION THROUGH BUNDLE ADJUSTMENT FEEDBACK-BASED ERROR DETECTION
SN - 978-989-674-028-3
AU - Hess-Flores M.
AU - A. Duchaineau M.
AU - J. Goldman M.
AU - I. Joy K.
PY - 2010
SP - 400
EP - 405
DO - 10.5220/0002816104000405