A novel hierarchical search motion estimation algorithm

A new hierarchical motion estimation algorithm that has better performance than other conventional hierarchical algorithms will be presented here. In this technique several sub-optimal algorithms are utilized without increasing the complexity of the circuit, at the same time the computation cost is reduced significantly while the accuracy is kept very close to the full search algorithm. The reduction of the computation is obtained by calculating the distortion between the decimated block of the current frame and that of the previous frame. The motion vector is found hierarchically by searching the candidate motion vectors found from previous level of search. Each higher level search uses partial distortion functions with decreased ratio of decimation. To reduce the computation further, only part of the locations in the search area is considered. The proposed algorithm is suitable for applications such as HDTV and MPEG-2 where motion estimation accuracy is important. It is also suitable for real-time applications such as video conference and H.263 where encoding speed is critical.