Efficient video motion estimation using dual-cross search algorithms

We present in this paper two new and efficient block-matching algorithms (BMAs) for fast motion estimation which is commonly used in motion-compensated video compression. The proposed algorithms are an improvement on existing BMAs and consist of three effective steps: (1) initial search center prediction; (2) early search termination; and (3) dual-cross search. Extensive simulation results and comparative analysis have shown that the proposed algorithms outperform conventional algorithms, like the three-step search, orthogonal search and diamond search, as well as the newly proposed algorithms, like the hexagonal search and adaptive rood pattern search, in terms of both peak signal-to-noise ratio and the number of search points evaluated.