Robust motion estimation for calibrated cameras from monocular image sequences

A new computational approach to estimate the ego-motion of a camera from se ts of point correspondences taken from a monocular image sequence is presen ted, The underlying theory is based on a decomposition of the complete set of model parameters into suitable subsets to be optimized separately; e.g., all stationary parameters concerning camera calibration are adjusted in ad vance (calibrated case). The first part of the paper is devoted to the desc ription of the mathematical model, the so-called conic error model. In cont rast to existing methods, the conic error model permits us to distinguish b etween feasible and nonfeasible image correspondences related to 3D object points in front of and behind the camera, respectively. Based on this "half -perspective" point of view a well-balanced objective function is derived t hat encourages the proper detection of mismatches and distinct relative mot ions. In the second part, some results of tests featuring natural image seq uences are presented and analyzed. The experimental study clearly shows tha t the numerical stability of the new approach is superior to that achieved by comparable methods in the calibrated case based on a "full-perspective" modeling and the related epipolar geometry, Accordingly, the accuracy of th e resulting ego-motion estimation turns out to be excellent, even without a ny further temporal filtering. (C) 1999 Academic Press.