SPATIOTEMPORAL APPROACH FOR TIME-VARYING GLOBAL IMAGE MOTION ESTIMATION

Image motion estimation using the spatiotemporal approach has largely relied on the constant velocity assumption, and thus becomes inappropr iate when the velocity of the imaged scene off the camera changes duri ng tile data acquisition time. Using a polynormal or a trigonometric p olynomial model for the time variation of the image motion, spatiotemp oral algorithms are developed in this paper to handle time-varying (bu t space-invariant) motion, Under these models, it is shown that time-v arying image motion estimation is equivalent to parameter estimation o f one-dimensional (1-D) polynomial phase. or phase-modulated signals, which allows one to exploit well-established results in radar signal p rocessing, When compared with alternative approaches, the resulting mo tion estimation algorithms produce more accurate estimates. Simulation results are provided to demonstrate the proposed schemes.