Feature-Based Sequence-to-Sequence Matching
Yaron Caspi, Denis Simakov and Michal Irani


Paper: "Feature-Based Sequence-to-Sequence Matching"
Contact us:
  Yaron   Denis   Michal
Our affiliation:
  Computer Vision Group
  Faculty of Mathematics and Computer Science
  Weizmann Institute of Science
Read abstract
See examples:

Large zoom difference.

  1. The setup: scene and cameras
  2. Two cameras (wide-angle and zoomed, with zoom difference 1:3) are placed next to each other and capturing the same scene.

  3. Input sequences
  4. Camera 1 (wide-angle) Camera 2 (zoomed)
    Large zoom difference: input from camera 1 Large zoom difference: input from camera 2
    Video: MPEG 2.75Mb Video: MPEG 1.8Mb

  5. Detect moving objects
  6. ... using background subtraction.

  7. Extract interest points
  8. ... by taking the centeroid of each blob, in each frame.

    Construct trajectories from these points

  9. Use the trajectories as features for matching algorithm
  10. Recover homography (spatial alignment) and time shift (temporal alignment)
  11. Fused images (wide-angle and zoomed)
    Large zoom difference: fused image
    Video: MPEG 2.75Mb

To the top of the page Last modified: Oct 06, 2006