2009. 10. 27. 14:40
Computer Vision
Harris' RAPiD
C. Harris and C. Stennett, “Rapid - a video rate object tracker,” in Proc 1st British Machine Vision Conference, Sep 1990, pp. 73–77.
ref.
C. Harris, “Tracking with rigid models,” in Active Vision, A. Blake and A. Yuille, Eds. MIT Press, 1992, pp. 59–73.
alpha-beta tracker
http://en.wikipedia.org/wiki/Alpha_beta_filter
Kalman filter
http://en.wikipedia.org/wiki/Kalman_filter
C. Harris and C. Stennett, “Rapid - a video rate object tracker,” in Proc 1st British Machine Vision Conference, Sep 1990, pp. 73–77.
bmvc-90-015.pdfref.
C. Harris, “Tracking with rigid models,” in Active Vision, A. Blake and A. Yuille, Eds. MIT Press, 1992, pp. 59–73.
RAPID (Real-time Attitude and Position Determination) is a real-time model-based tracking algorithm for a known three dimensional object executing arbitrary motion, and viewed by a single video-camera. The 3D object model consists of selected control points on high contrast edges, which can be surface markings, folds or profile edges.
The use of either an alpha-beta tracker or a Kalman filter permits large object motion to be tracked and produces more stable tracking results. The RAPID tracker runs at video-rate on a standard minicomputer equipped with an image capture board.
alpha-beta tracker
http://en.wikipedia.org/wiki/Alpha_beta_filter
Kalman filter
http://en.wikipedia.org/wiki/Kalman_filter