H.264运动估计算法UMHexagonS的斐波纳契数列优化

针对H.264运动估计算法UMHexagonS搜索步长和搜索模板中存在的使用固定搜索步长和搜索点冗余的不足,结合斐波那契数列和中心偏置特性对其进行改进。新算法使用斐波那契数列的递进关系确定UMHexagonS算法的搜索步长,其次删除UMHexagonS算法中存在计算冗余的搜索点,最后结合中心偏置特性对UMHexagonS算法的大六边形搜索模板进行了修改。实验结果表明,新算法在保持UMHexagonS算法的比特率和峰值信噪比(PSNR)的情况下缩短了运动估计时间,并且随着图像像素、图像复杂度和搜索范围的提高,运动估计时间越来越短。新算法在搜索范围为64的情况下,平均缩短了23.82%的运动估计时间。

Fibonacci optimized UMHexagonS algorithm for H.264 motion estimation

In order to overcome the shortcomings of using fixed search step and existing redundant search point in UMHexagonS algorithm of H.264 motion estimation,this paper combined the Fibonacci sequence with center-biased feature to improve it.Firstly,the search step was determined by the progressive relationship of the Fibonacci sequence.Secondly,some search points which lead to redundant computation were deleted.At last,the search template of big hexagon was modified by the center-biased feature.The experimental results show that the new algorithm maintains the bit rate and Peak Signal-to-Noise Ratio(PSNR) of UMHexagonS,and reduces the time of motion estimation.And with the improvement of image elements,image complexity and search range,the time for motion estimation becomes less and less,and it can be reduced by an average of 23.82% of the motion estimation time of UMHexagonS algorithm when the search range is 64.