Abstract: Based on motionlessness detection and motion vector prediction, an effective motion estimation algorithm combining local full search and down-sampling is proposed. It accelerates the search speed and maintains high search accuracy. The motion estimation algorithm is applied to the Frame Rate Up-conversion(FRUC) system. Experimental results show that the proposed algorithm not only reduces the calculation complexity, but also maintains satisfactory visual quality and PSNR.

Key words: motionlessness detection, motion vector prediction, down-sampling, Frame Rate Up-conversion(FRUC)

摘要： 基于静止检测、运动矢量预测，结合局部全搜索和下采样方法，提出一种有效的运动估计算法。将该算法用于基于运动估计运动补偿的帧频提升系统中。实验结果表明，该算法在减少计算量的同时，保证了内插帧有较高的主客观质量，能减少搜索点数，同时保持较高的搜索准确度。

关键词: 静止检测, 运动矢量预测, 下采样, 帧频提升

CLC Number: 

• TP391