|
Filtered Blending: A new, minimal Reconstruction Filter for Ghosting-Free Projective Texturing with Multiple Images
Martin Eisemann, Anita Sellent, Marcus Magnor
|
|
| Home | Team | Research | Teaching | Publications |
|
Martin Eisemann,
Anita Sellent,
and
Marcus Magnor:
"Filtered Blending: A new, minimal Reconstruction Filter for Ghosting-Free Projective Texturing with Multiple Images", in Proc. Vision, Modeling and Visualization (VMV), pp. 119–126, November 2007. Part of project "Floating Textures". [pdf] [bib] |
Whenever approximate 3D geometry is projectively texture-mapped from diffent directions simultaneously, annoyingly visible aliasing artifacts are the result. To prevent such ghosting in projective texturing and image-based rendering, we propose two dierent GPU-based rendering strategies: Filtered
blending and
Floating textures. Either approach is able to cope with imprecise 3D geometry as well as inexact camera calibration. Ghosting artifacts are effectively eliminated at real-time rendering frame rates on standard graphics hardware. With the proposed rendering techniques, better-quality
rendering results are obtained from fewer images, coarser 3D geometry, and less accurately calibrated images.

TU Braunschweig
- Fakultät für Mathematik und Informatik
- Computer Graphics
- Publications
- Filtered Blending: A new, minimal Reconstruction Filter for Ghosting-Free Projective Texturing with Multiple Images