In this paper, we present new 3D Amplitude-Modulation Frequency Modulation (AM-FM) methods for video image analysis. We develop a framework that allows us to reconstruct the input video using AM-FM representations and also provide motion estimates that correspond to these AM-FM representations. The proposed motion estimation method provides three motion estimation equations per channel filter (AM, FM motion equations and a continuity equation). The methods are implemented using efficient, separable 1D filterbanks. We demonstrate the method in motion tracking, trajectory estimation and video signal reconstruction. © 2008 IEEE.