1 / 16

Image-Based Rendering using Disparity Compensated Interpolation

Image-Based Rendering using Disparity Compensated Interpolation. EE362/PSYCH221 Class Project (Winter Quarter 2005-200 6) Aditya Mavlankar Information Systems Laboratory Stanford University. Outline. Virtual view synthesis using Image-Based Rendering (IBR) Brief survey of IBR techniques

bonita
Télécharger la présentation

Image-Based Rendering using Disparity Compensated Interpolation

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. Image-Based Rendering using Disparity Compensated Interpolation EE362/PSYCH221 Class Project (Winter Quarter 2005-2006) Aditya Mavlankar Information Systems Laboratory Stanford University

  2. Outline • Virtual view synthesis using Image-Based Rendering (IBR) • Brief survey of IBR techniques • Goal of the project • Disparity Compensated Interpolation (DCI) • Results • Summary

  3. Virtual View Synthesis Background Foreground IBR techniques generate novel views from input images Camera 1 Virtual camera (novel view) Camera 2

  4. Brief Survey of IBR Techniques • Classification of IBR techniques according to [1] • Rendering with explicit geometry • E.g. 3-D warping, Layered Depth Image (LDI) rendering, View-dependent texture mapping • Rendering with implicit geometry • E.g. View interpolation, View morphing • Rendering without geometry • E.g. Light field rendering, Lumigraph systems [1] H. –Y. Shum, S. B. Kang and S. –C. Chan, “Survey of image-based representations and compression techniques,” IEEE Transactions on Circuits and Systems for Video Technology, Vol. 13, No. 11, pp 1020-1037, Nov. 2003.

  5. Goal of the project • Come up with an IBR technique which • Requires no depth information, no correspondence information • Works well when disparity between two key views is not too high • Computational complexity does not depend on scene complexity • New view-point anywhere on the line joining the two camera centers • Generate video of view-point traversal in a static natural scene • See effect of inserting novel views on viewing experience • How many intermediate novel views required for smooth view-point traversal?

  6. Disparity Compensated Interpolation 0 5 0 4 1 3 2 View from Camera 1 Novel view, in the making View from Camera 2

  7. Design Parameters in Block-Matching • Block size • small: Spurious matches • big: Cannot adapt to detail of the scene • Image resolution • small: inaccurate matching • big: computational time • Which color channel(s) to use for matching? • trust luminance or chrominance?

  8. Results: Cartoon Results: Cartoon

  9. Results: Cartoon Results: Ballet (256x192), played at 8 fps

  10. Results: Cartoon Results: Ballet (256x192), played at 10 fps

  11. Results: Cartoon Results: Ballet (512x384), played at 10 fps

  12. Future Directions • Avoid spurious matches • Enforce some sort of continuity of disparity vectors • Object segmentation might help • Adaptation of block-sizes according to image content

  13. Conclusion • HVS perspective: Novel views are critical for smooth traversal of view-point • novel views: the more the better (provided the quality of intermediate novel views is not too bad) • An IBR technique was designed which obviates the need for complex geometry

  14. The End

  15. Results: Cartoon Results: Ballet (512x384), played at 1 fps

  16. Results: Cartoon Results: Ballet (256x192), played at 1 fps

More Related