MOTION ESTIMATION FROM SCALED ORTHOGRAPHIC PROJECTIONS WITHOUT CORRESPONDENCES

This paper presents two new eigenstructure-based algorithms for estima ting the motion parameters of a rigid object from scaled orthographic projections. The approach requires neither point correspondences betwe en time frames nor between projections at each time. 2D statistics fro m projections are utilized to recover 3D statistics, which are used fo r motion estimation. This results in significant speed-up compared to the approach where point correspondences need to be found. The first a lgorithm is suitable for general point configuration and requires thre e scaled orthographic projections at each time frame. The second algor ithm is suitable for planar configuration of data points and requires only two orthographic projections at each time instant. Results of sim ulated and real image experiments demonstrate the performance and effi ciency of the algorithms, as well as showing their limitations.