As the search range increases, most fast motion estimation algorithms can not keep the performance up close to the full search BMA. In this paper, we propose a new fast motion estimation method which can locate the small search area for a block of the current frame.
The basic idea is to obtain a small set of MV candidates and choose from them the one which satisfies the smallest MAD. To locate the proper set of MV candidates, the method uses spatial correlation of motion vectors as well as hierarchical search method.
The proposed method drastically reduces the number of matching operations to about 3.6% of that of the full search BMA, with the minor PSNR degradation of 0.25dB even in the worst case. Especially, the proposed method is suitable for VLSI realization. The excellent performance of the new algorithm is supported by extensive experimental results.
