Richard Szeliski and James Coughlan
The problem of image registration subsumes a number of problems and techniques in multiframe image analysis, including the computation of optic flow (general pixel-based motion), stereo correspondence, structure from motion, and feature tracking. We present a new registration algorithm based on spline representations of the displacement field which can be specialized to solve all of the above mentioned problems. In particular, we show how to compute local flow, global (parametric) flow, rigid flow resulting from camera egomotion, and multiframe versions of the above problems. Using a spline-based description of the flow removes the need for overlapping correlation windows, and produces an explicit measure of the correlation between adjacent flow estimates. We demonstrate our algorithm on multiframe image registration and the recovery of 3D projective scene geometry. We also provide results on a number of standard motion sequences.
|Published in||International Journal of Computer Vision|
All copyrights reserved by Springer 1997.