Title :
A Three-Frame Approach to Constraint-Consistent Motion Estimation
Author_Institution :
Visualization & Comput. Vision Lab, GE Global Res., v, NY
Abstract :
We study the motion estimation across three images by simultaneously enforcing the geometric constraint of trilinearity and the constant-brightness constraint. An iterative and hierarchical three-frame scheme is proposed to solve the geometry-consistent and appearance-consistent motion fields. It provides an uncalibrated approach to dense motion estimation, applicable to both 2D and 3D rigid scenes under perspective projection with unconstrained camera motion. Valid and consistent constraints are explored to alleviate the inherent aperture and over smoothing problems
Keywords :
image sequences; iterative methods; motion estimation; stereo image processing; video signal processing; 2D rigid scenes; 3D rigid scenes; appearance-consistent motion fields; constant-brightness constraint; constraint-consistent motion estimation; geometric constraint; geometry-consistent motion fields; hierarchical three-frame scheme; iterative scheme; perspective projection; trilinearity; unconstrained camera motion; Apertures; Cameras; Computer vision; Equations; Layout; Lighting; Motion estimation; Smoothing methods; Sun; Visualization;
Conference_Titel :
Pattern Recognition, 2006. ICPR 2006. 18th International Conference on
Conference_Location :
Hong Kong
Print_ISBN :
0-7695-2521-0
DOI :
10.1109/ICPR.2006.186
