UMHexagonS算法的研究与改进

Research and Improvement of UMHexagon S Algorithm

针对UMHexagon S算法存在的问题,提出了一种改进的UMHexagon S算法。算法充分利用运动矢量的中心偏置特性与运动同质性,采用递增数列步长的对称十字型搜索模板和多层次八边形搜索模板以减少无用的搜索点,根据运动矢量的变化趋势自适应地确定多层次八边形的搜索方向,用16点井字型搜索代替5×5全搜索,并改善了运动估计的粗定位。实验结果表明,与UMHexagon S算法相比,改进算法在保证PSNR和比特率基本不变的前提下,平均减少了18.53%的运动估计时间,尤其是对大运动视频序列的效果更好,最大可以减少25.17%的运动估计时间。

Aiming at the existing problem of UMHexagonS algorithm, an improved UMHexagonS algorithm is proposed in this paper. The algorithm makes full use of the center-biased characteristics of motion vector and motion homogeneity, symmetrical-cross search pattern and multi-octagon search pattern with increasing sequence step are used to reduce the useless search point. The searching direction of multi-octagon is adaptively determined ac- cording to the change trend of motion vector, and 5 x5 full search is replaced with 16 points numbersign search, and improves the coarse position of mo- tion estimation. The experimental results show that compared with UMHexagonS algorithm, the improved algorithm can reduce about 18.53% motion es- timation time with only negligible change of PSNR and bit rate, especially for high motion video sequences, it can reduce 25.17% motion estimation time.