Issue No. 03 - March (1997 vol. 19)
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/34.584105
<p><b>Abstract</b>—A method for computing the 3D camera motion (the <it>ego-motion</it>) in a static scene is described, where initially a detected 2D motion between two frames is used to align corresponding image regions. We prove that such a 2D registration removes all effects of camera rotation, even for those image regions that remain misaligned. The resulting <it>residual</it> parallax displacement field between the two region-aligned images is an <it>epipolar field</it> centered at the FOE (Focus-of-Expansion). The 3D camera translation is recovered from the epipolar field. The 3D camera rotation is recovered from the computed 3D translation and the detected 2D motion. The decomposition of image motion into a <it>2D parametric motion</it> and <it>residual epipolar parallax displacements</it> avoids many of the inherent ambiguities and instabilities associated with decomposing the image motion into its <it>rotational</it> and <it>translational</it> components, and hence makes the computation of ego-motion or 3D structure estimation more robust.</p>
Motion analysis, ego motion, video stabilization, plane-plus-parallax.
S. Peleg, M. Irani and B. Rousso, "Recovery of Ego-Motion Using Region Alignment," in IEEE Transactions on Pattern Analysis & Machine Intelligence, vol. 19, no. , pp. 268-272, 1997.