针对复杂剧烈运动视频信号在进行运动估计时出现的预测误差问题,提出基于运动矢量预测性程度的快速运动估计算法。根据局部运动矢量的一致性程度与运动矢量在时间上相关性程度的关系,决定运动矢量的可预测性程度,将待估计的运动矢量分类为不可预测、中度可预测和高度可预测,不同的可预测性采用不同的搜索策略进行运动估计。实验结果表明,该算法能提高运动估计的速度,并且具有与其他快速运动估计算法近似或更好的PSNR性能。
In this paper, an effective classifiable prediction mechanism according to the degree of motion vector prediction is suggested for more fast motion estimation, which classifies the motion vectors as no predictable motion vector, medial predictable motion vector or high predictable motion vector, then different proposed search schemes are adopted for motion estimation. Experimental results show the proposed algorithm reduces the prediction error and has a significant computational speedup compared with other algorithms, but offers a similar, even better performance.