Estimating motion parameters and spatial position of object is very important in target tracking and computer vision. The constraint for obtaining 3D motion and coordinates from a monocular image sequence is first presented and proved in this paper. With this contraint，a nonlinear recusive filtering algorithm for estimation 3D motion parameters and 3D spatial position of a point target from monocular images has been proposed.The performance of the proposed technique on simulated image sequence is also involved.