摘要
针对H.264视频压缩中模式决策部分计算复杂度过高的问题,提出了一种实时的快速模式决策算法。首先,在帕累托最优宏块分类方法的基础上制定类决策指标,划分宏块类;然后,检测宏块的运动活跃性,为当前宏块选择率失真代价较小的模式,提前作出模式决策,得到最佳模式。实验结果表明,该算法与参考编码器相比,在比特率平均增加0.09%、峰值信噪比平均损失0.19dB的情况下,编码时间平均减少了64.28%。
With the problem of high computation complexity in mode dicision pattern of H.264 video compression,a real-time fast mode decision algorithm was poposed. Firstly, the method formulated class decision metrics, divided MB class based on Pareto-optimal macroblock （MB） classification scheme; Then, the method detected the motion activity of macroblocks,selected mode which with low rate-distortion cost for current macroblock,made mode dicision in advance and got the best mode.Compared with the reference coder, experimental results show that the proposed algorithm can achieve a reduction of 64.28% encoding time on average,with a negligible average 0.09% bit rate increasing and a PSNR（Peak Signal to Noise Ratio） loss of 0.19dB.
出处
《电子技术应用》
北大核心
2013年第4期109-111,114,共4页
Application of Electronic Technique
