Motion estimation algorithm based on cross-correlation and pyramidal decomposition

One of the challenges in practical motion estimation is how to obtain high accuracy with sufficient robustness to both brightness variation and additive noise. Motivated by the facts that the normalized spatial cross-correlation is invariant to brightness and the pyramidal decomposition technique can easily realize large-displacement estimation meanwhile it is immune to noise, we introduce a motion estimation algorithm which employs the pyramidal decomposition technique to decompose the image into several layers, ensuring the displacement in each image layer is small, at the same time weakening the impact of noise on estimation accuracy to some degree, then the cross-correlation function is used to estimate the displacement of each image layer, at last the displacement of each image layer is weighted to obtain the whole displacement. Simulated and real experiments demonstrate that the algorithm proposed in this paper is suitable for large and small displacement, and it has high accuracy even for the noisy and brightness changing image sequence.