基于H．264自适应阈值的快速运动估计算法

A Fast Motion-estimation Algorithm Based on Adaptive Threshold for H.264

块匹配运动估计在视频编码中有着举足轻重的地位,全搜索算法是最好的搜索算法,但其计算量是最大的,所以提出一种H.264自适应阈值的快速运动估计算法。该算法是利用相邻块的运动矢量来预测初始搜索点,并根据一定的准则来确定静止块,减少了搜索的冗余度,再用基于梯度的十字优先菱形算法来进行搜索。与DS菱形算法相比,该算法具有更高的灵活性,能实时的提前退出搜索,并且在搜索点数上明显减少,且没减少搜索精度。

Block matching motion estimation (BMME)plays a very important role in video coding. Full search(FS)is the optimal algorithm for BMME, but its tremendous computation is very large.So this paper presents an adaptive threshold fast motion-estimation algorithm based on H.264 Taking advantage of motion-vectors of the nearby macro- blocks to find the original searching point and ascertain the slack block on the base of definite rules,it decrease the redundancy,at last,it will search the best block using cross priority diamond search arithmetic based on grades.Compared with DS diamond algorithm,this algorithm is more flexible and can exit search in advance at the real time, also decrease the search points clearly and don't reduce the search precision.