H.264中一种基于搜索范围自适应调整的运动估计算法

摘要

视频编码标准 H.264中的固定搜索范围设定，使得运动估计单元的计算量和存储读写带宽过大，给实时编码带来了困难。该文提出一种基于搜索范围自适应调整的运动估计算法。算法首先根据当前编码块预测运动矢量(PMV)点的绝对误差和(SAD)以及相邻已编码块的信息，判断 PMV 的预测准确程度，自适应地确定搜索范围(SR)大小；然后通过检测编码块所在区域的运动方向特性，确定 SR 内的有效搜索方向。仿真结果表明：相比于全搜索算法和 UMHexagon 算法，该文算法分别能够平均节省91%和18%左右的运动估计时间，而码率和图像 PSNR 基本保持不变。%To reduce the computational complexity and memory access bandwidth of motion estimation in H.264 video code, an algorithm to decide the search range for motion estimation is proposed. Firstly, the Search Range (SR) is adaptively selected based on the accuracy of Predicted Motion Vector (PMV), which is measured by using the Sum of Absolute Difference (SAD) of neighboring blocks. Then, the motion activity of coding region is also considered for search direction selection in the search area. The extensive experimental results show that the proposed algorithm is able to reduce 91% and 18% of the computation time on average compared to full search and UMHexagon algorithm, respectively, with almost the same bit rate and PSNR.