Automatic feature-based global motion estimation in video sequences

In this paper, automatic and efficient global motion estimation for video sequences is proposed. In this estimation, we find and pick up pairs of feature points in the edge lines within the successive frames using the match algorithm. We then choose the best set of three pair points to measure the parameters of the Affine Model which describes the global motion. The result shows that we can obtain the Affine parameters of global motion efficiently in video sequences.