【摘要】 在基于搜索起点预测和梯度式菱形搜索方法的基础上,提出了一种运动矢量场自适应搜索算法(MVFSAS)。实验结果表明,该算法以较小的搜索代价取得了与全搜索相当的效果,并在搜索速度和搜索效果方面优于三步(3SS)和四步（4SS)快速运动搜索算法。更多还原

【Abstract】 Based on prediction of searching starting-point and gradient diamond-shaped searching technique, the paper proposes a motion vector field self-adaptive searching algorithm(MVFSAS). Experiments show that the algorithm can obtain almost the same PSNR as full search (FS) at very low computation cost, and it outperforms both three-step search(3SS) and four-step search(4SS) algorithm.更多还原