An hierarchical motion estimation method using adaptive image down-sizing
Hyungjun Lim, Tae Gyoung Ahn · 2014
This paper proposes a new hierarchical motion estimation method. Typical hierarchical motion estimation method usually down-sizes video frames by using a fixed sampling position, whereas the proposed method uses a non-fixed sampling position. In the proposed method, the sampling position of the current frame is determined by the motion estimation results of previous frames. Experimental results demonstrate that the proposed method obtains more accurate motion vectors than the typical hierarchical motion estimation method.