带有阀值的基于节点搜索的可变形块运动估计算法

Motion Estimation Algorithm by Node Search Based On Deformable Block Matching With Threshold

文章在讨论基于节点位移的可变形块运动模型的基础上,分析了NS-DBMA的搜索过程和流程,从而提出了带有阀值的基于节点搜索的可变形块运动估计算法(TNS-DBMA)。以四节点模型为例,详细研究了该算法的流程,并利用两个序列进行测试,测试结果表明,TNS-DBMA算法的预测图像效果与NS-DBMA算法接近,但它的运算量大大的减少,比NS-DBMA算法减少一半多。

This paper first discusses the deformable block model based on the displacement nodes, and analyzes the search process and flow of NS-DBMA algorithm, then proposes the motion estimation algorithm by node search based on deformable block matching with threshold. Four-node model as an example, the paper further studies the technical flow of the algorithm, and the test by two sequences is done, the test results show that forecast images effect of TNS DBMA algorithm is close to NS-DBMA algorithm, and however, the TNS-DBMA algorithm greatly reduces the computational complexity, at least to half as much as that the NS-DBMA algorithm does.