A hierarchical motion estimation algorithm using nonlinear pyramid for MPEG-2

In this paper, we introduce a new block-based hierarchical motion estimation algorithm using nonlinear pyramid (HMENP). The proposed algorithm based on nonlinear pyramid reduces the computational complexity dramatically since the implementation of nonlinear pyramid is multiplication free and addition-free. The variable block size was used in HMENP to avoid the drawback of the constant block size HMENP in describing small object motion activities. The proposed motion estimation algorithm has been implemented and tested in an MPEG-2 encoder. It is shown that the HMENP has a good performance in terms of the peak-to-peak signal-to-noise ratio as well as the subjective quality.