Estimation of structure and six-degree-of-freedom motion of maneuvering objects through measurements of feature positions in long, multiple- camera image sequences is widely recognized to have broad industrial military and space applications, particularly in the control of autonomous systems. This report focuses on the case of maneuvering objectives thereby removing restrictive assumptions concerning the mode of translational and rotational motion which are commonly employed in many existing methods. Object maneuvers, being' smooth and time correlated, are modelled as first-order Gauss-Markov processes for both translational and rotational motion. In the literature, rotational motion is often parameterized with unit quaternions even though constrains on the quaternion norm are not easily enforced, roll-pitch-yaw parameterizations have been reported to be poorly behaved and have led to computationally demanding implementations, and results using the Euler angle- axis parameterization in recursive motion and structure estimation are not available. In this report, six-degree-of-freedom, nonlinear, approximate state estimation filters for quaternion, roll-pitch-yaw and angle-axis parameterizations are compared in terms of estimation performance for maneuvering object trajectories. Special consideration is given to the problem of imposing unit norm on the estimated quaternion since previously proposed methods led to filter instability, particularly in angular velocity estimation. Motion, Motion analysis, Structure fron motion, Maneuvering objects, Target tracking, Kalman filtering, Extended Kalman filtering, Quaternions, Angle axis, Roll pitch yaw, Rotational motion, Stereo, Image sequences.