论文部分内容阅读
为了降低视频编码器的运算负荷，提出一个空时相关多分辨半像素精度块匹配运动估计算法，这个算法利用运动矢量场的空间和时间相关性及多层结构中父子块间的相关性，缩小局部运动搜索范围，得到低运算复杂性。利用一个自适应门限调整算法，使运动搜索次数对不同视频序列近于一致。模拟结果表明，该算法运动补偿增益和运动矢量熵也有明显改进，搜索次数低于著名的三步法。最后讨论了半像素精度和变块大小运动估计方案。
In order to reduce the computational load of video encoder, a space-time correlated multiresolution half-pel block matching motion estimation algorithm is proposed. This algorithm uses the spatial and temporal correlation of motion vector field and the correlation between parent-child blocks in multi-layer structure Local motion search range, get low computational complexity. Using an adaptive threshold adjustment algorithm, the number of motion search for different video sequences is almost the same. The simulation results show that the algorithm also significantly improves the motion compensation gain and motion vector entropy, and the search times are lower than the famous three-step method. Finally, the half-pixel accuracy and variable block size motion estimation scheme are discussed.