摘要
本文提出了一种预测质量可控的快速运动估计搜索算法 .该算法利用运动矢量的空间、时间相关性 ,通过预测初始搜索中心位置、判定是否为静止宏块、调整搜索窗口、预测质量可控的渐进式块匹配准则搜索等步骤进行运动估计 ,并引入调节函数在预测质量和搜索速度之间增加了可控性 .实验结果表明本算法大大减少了计算量 ,明显提高了运算速度 ,且具有很好的重建视频质量 。
This paper presents a fast motion estimation search algorithm with adjustable predictive quality.Taking advantage of the spatial and temporal correlation of motion vectors,motion estimation of the algorithm is realized by predicting initial search center location,analyzing whether a silent macroblock or not,adjusting windows size,and using progressive block matching measure of adjustable predictive quality,and at the same time,introducing adjustable function to add controllability between predictive quality and searching speed.Experimental results show that the algorithm can reduce computation amount significantly,increase computing speed greatly,provide good reconstructed video quality,and be applied for low bit rate video encoding.
出处
《电子学报》
EI
CAS
CSCD
北大核心
2004年第6期980-982,共3页
Acta Electronica Sinica
基金
国家自然科学基金 (No .60 0 72 0 39)
关键词
预测质量可控
渐进式块匹配准则
运动估计
视频编码
adjustable predictive quality
progressive block matching measure
motion estimation
video encoding
