Advanced search
1 file | 562.95 KB

Variational multi-image stereo matching

Simon Donné (UGent) , Bart Goossens (UGent) , Jan Aelterman (UGent) and Wilfried Philips (UGent)
Author
Organization
Abstract
In two-view stereo matching, the disparity of occluded pixels cannot accurately be estimated directly: it needs to be inferred through, e.g., regularisation. When capturing scenes using a plenoptic camera or a camera dolly on a track, more than two input images are available, and - contrary to the two-view case -pixels in the central view will only very rarely be occluded in all of the other views. By explicitly handling occlusions, we can limit the depth estimation of pixel (P) over right arrow to only use those cameras that actually observe (p) over right arrow. We do this by extending variational stereo matching to multiple views, and by explicitly handling occlusion on a view-by-view basis. Resulting depth maps are illustrated to be sharper and less noisy than typical recent techniques working on light fields.
Keywords
stereo, light field, depth estimation

Downloads

  • 388-A8MV-141.pdf
    • full text
    • |
    • open access
    • |
    • PDF
    • |
    • 562.95 KB

Citation

Please use this url to cite or link to this publication:

Chicago
Donné, Simon, Bart Goossens, Jan Aelterman, and Wilfried Philips. 2015. “Variational Multi-image Stereo Matching.” In IEEE International Conference on Image Processing ICIP, 897–901. IEEE.
APA
Donné, S., Goossens, B., Aelterman, J., & Philips, W. (2015). Variational multi-image stereo matching. IEEE International Conference on Image Processing ICIP (pp. 897–901). Presented at the IEEE International Conference on Image Processing (ICIP), IEEE.
Vancouver
1.
Donné S, Goossens B, Aelterman J, Philips W. Variational multi-image stereo matching. IEEE International Conference on Image Processing ICIP. IEEE; 2015. p. 897–901.
MLA
Donné, Simon, Bart Goossens, Jan Aelterman, et al. “Variational Multi-image Stereo Matching.” IEEE International Conference on Image Processing ICIP. IEEE, 2015. 897–901. Print.
@inproceedings{6848469,
  abstract     = {In two-view stereo matching, the disparity of occluded pixels cannot accurately be estimated directly: it needs to be inferred through, e.g., regularisation. When capturing scenes using a plenoptic camera or a camera dolly on a track, more than two input images are available, and - contrary to the two-view case -pixels in the central view will only very rarely be occluded in all of the other views. By explicitly handling occlusions, we can limit the depth estimation of pixel (P) over right arrow to only use those cameras that actually observe (p) over right arrow. We do this by extending variational stereo matching to multiple views, and by explicitly handling occlusion on a view-by-view basis. Resulting depth maps are illustrated to be sharper and less noisy than typical recent techniques working on light fields.},
  author       = {Donn{\'e}, Simon and Goossens, Bart and Aelterman, Jan and Philips, Wilfried},
  booktitle    = {IEEE International Conference on Image Processing ICIP},
  isbn         = {978-1-4799-8339-1},
  issn         = {1522-4880},
  language     = {eng},
  location     = {Quebec, Canada},
  pages        = {897--901},
  publisher    = {IEEE},
  title        = {Variational multi-image stereo matching},
  url          = {http://dx.doi.org/10.1109/ICIP.2015.7350929},
  year         = {2015},
}

Altmetric
View in Altmetric
Web of Science
Times cited: