论文部分内容阅读
为减少视频编码标准 H.2 6 4中树状结构运动估计的高运算复杂度 ,将连续消除算法 (SEA )应用于树状结构的运动估计中。在性能完全相同的条件下 ,与 H.2 6 4整像素运动矢量全搜索算法相比 ,其块匹配运算量减少到原来的 1%～ 2 0 %。利用不同大小块之间因部分重叠 ,其运动矢量具有更强相关性的特点。提出了一种简单有效地确定运动矢量搜索初值的方法 ,并应用两种快速的目标运动矢量判定方法 ,改进了 SEA算法。在性能损失可以忽略的情况下 ,进一步将基本 SEA运算量降低到全搜索块匹配运算量的 0 .3%～4 %。
To reduce the computational complexity of tree structure motion estimation in video coding standard H.264, a continuous elimination algorithm (SEA) is applied to the tree structure motion estimation. Compared with the full search algorithm of H.264 integer pixel motion vector, the block matching operation reduces to 1% ~ 20% under the same performance. The use of different size between the blocks due to partial overlap, the motion vector has a stronger correlation characteristics. A simple and effective method to determine the initial value of motion vector search is proposed. Two fast target motion vector determination methods are applied to improve the SEA algorithm. In the case of negligible performance loss, the basic SEA computation is further reduced to 0.3% ~ 4% of the full search block matching computation.