A 3-dimensional hybrid search algorithm for fast block matching in video coding

Motion estimation is known to be one of the most important parts of real-time video conferencing application. A 3-dimentional hybrid search algorithm for fast block matching motion estimation is proposed in this paper. The algorithm not only uses information of the neighbour motion vectors, but also motion vectors from the previous frame. A simple classification method is proposed to categorise blocks as motion or non-motion. A different search method is used for each type of block. Experimental results demonstrate the proposed method achieves better performance and requires the least computation among existing algorithms.