1 / 16

Robust Wide Baseline Stereo from Maximally Stable Extremal Region

Robust Wide Baseline Stereo from Maximally Stable Extremal Region. J. Matas, O Chum, M. Urban, T. Pajdle BMVC 2002. Introduction. Objective : Finding correspondences in two images. An enabling step toward many applications. Distinguished Regions. Maximally Stable Extremal Regions.

hoang
Download Presentation

Robust Wide Baseline Stereo from Maximally Stable Extremal Region

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. Robust Wide Baseline Stereo from Maximally Stable Extremal Region J. Matas, O Chum, M. Urban, T. Pajdle BMVC 2002

  2. Introduction Objective: Finding correspondences in two images. • An enabling step toward many applications.

  3. Distinguished Regions

  4. Maximally Stable Extremal Regions Distinguished Regions • Stability • Adjacency preserving • Invariance to affine • Multi-scale detection

  5. Extremal/Maximal Regions g=0.2 g=0.4 g=0.9 Definition: A set of all connected components (pixels) below all thresholds.

  6. Extremal/Minimal Regions g=0.2 g=0.4 g=0.9 Definition: A set of all connected components (pixels) above all thresholds.

  7. Maximally Stable Extremal Regions Stable Regions: An extremal region stays virtually unchanged over a range threshold.

  8. Maximally Stable Extremal Regions

  9. Maximally Stable Extremal Regions

  10. Maximally Stable Extremal Regions Descriptor Location of intensity maximum/minimum, Threshold. Measurement region Ellipse, circle, rectangular image patches, contours. Similarity Mohalanobis distance, correlation, etc.

  11. Epipolar Geometry (EG) Saliency Detection Region Matching Rough Affine Refine EG

  12. Applications Estimation of Epipolar Geometry

  13. Applications Tracking of license plates

  14. Applications Face tracking Pixels of color image are ordered by Mahananobis distance to the estimated skin-tone Gaussian distribution in R-G space.

  15. Remarks • Stable salient point/regions detection. • Application in epipolar geometry and tracking. • Potential feature transform descriptor.

  16. Thank you

More Related