Proceedings PaperMotion From A Sequence Of Images
|Format||Member Price||Non-Member Price|
We propose a method for the recovery of motion parameters of a rigid object moving through environment with constant but arbitrary linear and angular velocities. The method uses temporal information from a sequence of images such as those taken by a mobile robot. Spatial information contained in the images is also used. The temporal sequence, combined with the assumption of constant velocities, provides powerful constraints for the motion trajectory of rigid objects. We derive a closed form solution for the rigid object trajectory by integrating the differential equations describing the motion of a point on the tracked object. The integrated equations are non-linear only in angular velocity and are linear in all other motion parameters. These equations allow the use of a simple least-square error minimization criterion during an iterative search for the motion parameters. Experimental results demonstrate the power of our method in fast and reliable convergence.