中文摘要：

针对现有的降分辨率视频转码运动矢量合成算法没有考虑输入运动矢量的精确度，从而使得合成误差较高的问题，提出了一种基于精确度的降分辨率视频转码运动矢量合成算法。该算法分为两次合成，首先基于绝对差值和（SAD）对转码输入流中的运动矢量进行平均合成，然后再基于率失真函数，对转码输出流中的相邻运动矢量和第1次合成得到的运动矢量进行选择合成。仿真结果表明，与现有算法相比，该算法在保持较低计算复杂度的情况下，显著降低了运动矢量合成的误差。

英文摘要：

The current motion vector composition algorithms for reduced spatial resolution video transcoding does not consider the accuracy of input motion vectors. So the errors of these algorithms are rather big. In order to solve this problem, this paper proposes a motion vector composition algorithm hased on the accuracy for reduced spatial resolution video transcoding. This algorithm includes two compositions. Firstly, it composes the input motion vectors by averaging them based on SAD. Secondly, it composes the neighboring motion vectors and composed motion vector in the first composition by selecting them based on rate distortion function. The simulation results show that the proposed algorithm reduces the error compared with the algorithms while maintaining the low complexity.