摘要
在H.S.Kang的ME Skip运动估计算法的基础上,提出了一种改进的ME Skip运动估计算法.新算法对运动估计匹配误差差分绝对值之和(SAD)的阈值设置作了改进.新算法设置了两个SAD阈值,较小的SAD阈值用以克服误差传播,较大的SAD阈值用以尽可能的避免运动估计.当基于预测运动矢量的匹配误差介于两SAD阈值之间时,用小窗口搜索代替运动估计.实验结果表明在编码效率略有提高的同时,新算法进一步减少了运动估计运算量的10%～20%.
Based on the ME-Skip algorithm for motion estimation proposed by H.S.Kang, an advanced algorithm for motion estimation is presented. An improvement is made on the setting of matching error (measured by SAD) thresholds. There are two SAD thresholds in the new algorithm. The smaller one is used to overcome the propagation of the matching error, while the larger one is applied to avoid the execution of ME to the greatest degree. When the matching error based on the prediction of the motion vector is between two SAD thresholds, a search restricted in a small window is used to replace the original motion estimation. Results show that with a little increase in coding efficiency, the proposed algorithm reduces computational complexity in motion estimation by as much as 10%-20%.
出处
《西安电子科技大学学报》
EI
CAS
CSCD
北大核心
2005年第1期108-111,共4页
Journal of Xidian University
基金
国家自然科学基金资助项目(60002007)
关键词
MPEG-4
视频压缩
运动估计补偿
Algorithms
Computer simulation
Image compression
Motion compensation
