6 Degree-of-Freedom Motion Estimation of a Moving Target using Monocular Image Sequences

A novel estimation method for the 6 degree-of-freedom (6-DOF) motion of a target moving with slowly varying velocity is proposed, which is based on the monocular image sequences from a moving camera. The homography matrix for moving targets is derived. It is shown that the dynamic model of the target in 6-DOF motion can be described as a linear time-varying system. Consequently, the Kalman filter can be directly applied. Our estimation method does not need a priori information of target depth or size. The observability condition of our estimator is much milder than that of the previously known bearings-only tracking methods.