Abstract: A novel adaptive motion estimation algorithm based on motion characteristics is proposed in this paper.Two approaches are adopted to improve the estimation performance.First,an adaptive model for starting point prediction is presented.Based on the variation of the correlation between the adjacent blocks,the parameters of the model can be modified correspondingly to minish the difference between the prediction result and the best motion vector.Second,an alterable search pattern is introduced to improve the searching efficiency by adjusting its size and shape according to the motion characteristics of the blocks.Experiment results show that the proposed algorithm greatly outperforms other popular fast motion estimation algorithms in terms of both PSNR and search speed.

摘要： 提出了一种新的基于运动特征的自适应运动估计算法。该算法主要基于两方面：（1）建立具有自适应特性的搜索起点预测模型,根据运动相关性的变化调整模型参数,使预测结果更加接近最佳运动矢量。（2）采用的搜索模板可以根据物体的运动特征调整大小和形状,从而提高搜索效率。实验结果表明,该算法在PSNR和搜索速度两方面均明显优于常用的快速算法。