Abstract: Based on the classic Three Step Search(TSS) algorithm, a more efficient motion estimation method, Double Tracing Three Step algorithm (DTTSS) is proposed. After the first step, DTTSS selects two best match points as start points in the next step for double tracking. With relatively high search speed, DTTSS can avoid the local optimum problem in the TSS. This algorithm is compared with Full Search(FS), Local Full Search(LFS), Three Step Search(TSS) in Motion Estimation and Motion Compensation(ME/MC) frame rate up conversion system. Experimental results show the effectiveness in the average number of search points of each block, the accuracy, the visual effect of the interpolated frame and its PSNR.

Key words: double-tracing, motion estimation, motion compensation, frame rate up

摘要： 基于经典的三步法(TSS)，提出一种更有效的运动估计方法——双重跟踪三步法(DTTSS)。DTTSS在TSS第1步搜索结束后选取2个最优匹配点为下一步的起始点进行双重跟踪，在保持较高搜索速度的前提下，DTTSS可有效避免TSS中搜索陷入局部最优的问题。将该算法应用于运动估计/运动补偿的帧频提升系统。将实验结果与全搜索法、局部全搜索法、三步法进行对比分析，根据平均每块搜索点数、准确度、内插帧视觉效果及其峰值信噪比验证该算法的有效性。

关键词: 双重跟踪, 运动估计, 运动补偿, 帧频提升

CLC Number: 

• TP18