基于边界约束的样品视频非对称运动估计方法

针对常规运动估计方法应用到样品视频编码时存在无效搜索点冗余搜索、有效搜索点遗漏的问题,提出基于边界约束的非对称运动估计方法。首先,在原有预定搜索范围的基础上,采用定点统计运动参量的方法对科学仪器的样品视频运动性能进行测试统计,根据统计结果,对视频运动范围设定边界,减少运动搜索点数;然后提出了基于边界约束的非对称搜索模型,依据样品视频的运动特征,优化搜索算法。来自电子探针和电子显微镜的不同样品视频编码实验表明,与多方向搜索算法(MDS)比较,所提方法的运动估计时间缩短了约33%,编码性能保持甚至超过了多方向搜索算法。

Asymmetrical sample video motion estimation with boundary constraint

When applying the conventional motion estimation algorithms for sample video coding,the invalid search points will be searched redundantly and some valid points will be missed,So an asymmetrical motion estimation method with boundary constraint is proposed.Firstly,on the basis of original search scope,the motion performance of sample video in the instrument is measured and recorded with statistical method of motion parameters at fixed points.According to the statistical results,a motion boundary is set to reduce the search points.And then,in accordance with the motion feature of sample video,an asymmetrical search model with boundary constraint is brought out to optimize the search algorithm.The different sample videos from electron probe and electron microscope are employed to test the performance of the proposed estimation method.It proves that the method compresses the motion estimation time by about 33% compared with the multi-direction search(MDS) algorithm.And in terms of coding quality,it maintains and even exceeds the MDS method.