1 / 19

Efficient Large-Scale Stereo Matching ACCV 2010

Efficient Large-Scale Stereo Matching ACCV 2010. Andreas Geiger, Martin Roser , Raquel Urtasun ,. M.S. Student, Hee -Jong Hong 04. 15. 2014. Outline. Introduction Related Works Proposed Method Experimental Results Conclusion. Introduction. Motivation

daxia
Télécharger la présentation

Efficient Large-Scale Stereo Matching ACCV 2010

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Efficient Large-Scale Stereo MatchingACCV 2010 Andreas Geiger, Martin Roser, Raquel Urtasun, M.S. Student, Hee-Jong Hong 04. 15. 2014

  2. Outline • Introduction • RelatedWorks • Proposed Method • ExperimentalResults • Conclusion

  3. Introduction • Motivation • On Vehicle system -> Platform needs • Outdoor Scene • Illumination change • High resolution image • Wide slanted region • …

  4. Introduction • Stereo Problems • Ambiguities • Textureless regions • Exposure • Non-Lambertian surfaces • Δz grows quadratically

  5. Related Works • Local Method • Winner Takes All • Global Method • Minimize 1D/2D energy • Seed & Grow Method • Grow disparity components from random seeds

  6. Proposed Method

  7. Idea • Assumption: Rectified images • Match Point on epipolar line

  8. Idea • Image pairs contain ’easy’ and ’hard’ correspondences

  9. Idea • Robustly match ’easy’ correspondences on regular grid • Down sampling computation -> fast • Named : Support Point

  10. Idea • Build prior on dense search space • dense matching

  11. ELAS • Notation • Robust Support Points • Disparity • Observation • Local image features

  12. ELAS • Algorithm • Split image domain into support points and dense pixel • Assume factorization of distribution over disparity • Graphical Model

  13. ELAS • Proposed Prior (1) (2)

  14. ELAS • Proposed Likelihood • Laplace distribution (3) Left Image Right Image

  15. Experimental Result

  16. ExperimentalResults • Middlebury Benchmark

  17. Joint Histogram Based Cost Aggregation for Stereo Matching - TPAMI 2013 Conclusion

  18. Joint Histogram Based Cost Aggregation for Stereo Matching - TPAMI 2013 Conclusion • Contribution • Simple prior based on sparse feature matches • Reduced ambiguities and run-time • Takes into account slanted surfaces • Real-time 3D reconstruction of static scenes on CPU • C++ / MATLAB code available at http://cvlibs.net • Futurework • Develop better priors • Incorporate segmentation / global reasoning on lines • GPU implementation • (goal: 20 fps at 1-2 megapixels) • Employ as unitary potentials on global methods => smaller label sets

  19. Thank you!

More Related