| A. Bartoli, N. Dalal, B. Bose, and R. Horaud. From video sequences to motion panoramas. In IEEE Workshop on Motion and Video Computing, December 2002. |
....in zoom and focus. This type of camera motion is conveniently described by a plane to plane projective transformation and can be estimated from image point correspondences without any prior knowledge. Cameras mounted on tripod (as used in athletic events) satisfy these conditons to a good accuracy [1]. Second, we assume that image regions corresponding to individual body parts can be tracked through the video sequence. Third, we assume that the focal length varies in a way such that the apparent image size of the object of interest (here a human body) remains constant. The latter implies that ....
....denotes respective positions in j th frame. 2.1 Camera Motion Estimation We assume that camera s centre of projection is fixed and only focal length f , tilt angle 1 and pan angle 2 vary. These assumptions are valid for cameras mounted on tripods and hence sufficient for most video sequences [1]. Let H be the plane plane projective transformation or homography warping the (j 1) th frame to current j th frame. At each time instant j, we estimate homographies H parameterized over f , 1 and 2 by minimizing dense photometric information between two consecutive frames [1, 11] and ....
[Article contains additional citation context not shown here]
A. Bartoli, N. Dalal, B. Bose, and R. Horaud. From video sequences to motion panorama. In Proceedings of the IEEE workshop on Motion and Video Computing, Orlando, Florida, USA, 2002.
No context found.
A. Bartoli, N. Dalal, B. Bose, and R. Horaud. From video sequences to motion panoramas. In IEEE Workshop on Motion and Video Computing, December 2002.
Online articles have much greater impact More about CiteSeer.IST Add search form to your site Submit documents Feedback
CiteSeer.IST - Copyright Penn State and NEC