基于H.264运动估计的空间预测算法

Spatial Prediction Algorithm Based on Motion Estimation in H.264

提出一种新的空间自适应预测算法,将其与H.264运动估计中的非对称十字型多层次六边形格点(UMH)搜索算法相结合,预测当前块的运动趋势。实验结果表明,在峰值信噪比和码率几乎无变化的情况下,该算法相比UMH算法,运动估计时间平均减少了9%,运动搜索点数平均减少了13%,运动估计的复杂度显著降低,编码器的实时性得到增强。

This paper proposes a spatial-adaptive search algorithm, and combines it with Unsymmetrical Multi-resolution Hexagon（UMH） search algorithm in H.264 to predict the motion trend of present block. Experimental results show that the algorithm can decrease 9% motion estimation time and 13% search points without losing PSNR and bit rate, compared with UMH algorithm. The computational complexity of motion estimation is reduced evidently and real-time characteristic is strengthened.