高效的H.264时域误码掩盖算法

Efficient temporal error concealment algorithm for H.264

为提高在视频通信中重建图像质量,针对最新 H.264视频编码标准的特点,提出了一种基于分数像素边界匹配搜索的时域误码掩盖算法.宏块丢失导致的解码错误会在时间和空间上扩散,使得重建图像质量严重下降.该算法充分利用了运动矢量的空间相关性、H.264标准中运动矢量的精度以及边界像素信息,更精确地恢复丢失宏块的运动矢量,从而提高所恢复视频信号的质量.仿真结果表明:对于不同运动程度序列和不同宏块丢失率,该算法优于传统的整像素的边界匹配算法,具有很强的鲁棒性,当应用于较高的宏块丢失率及运动较复杂的序列时,所恢复视频序列的亮度信号峰值信噪比相对传统算法提高了1.0～2.5dB.

Decoding errors in video communications caused by macroblock loss will propagate in both the temporal and spatial domains and will substantially reduce the video quality. The quality of reconstructed images can be enhanced by a temporal error concealment algorithm based on a fractional-pixel boundary-matching search for the newest international video coding standard H. 284/AVC. The algorithm efficiently exploits the spatial correlation and the high precision of the motion vectors in H.264 to more accurately recover the lost motion vectors and improve the video quality. For video sequences with rapidly changing motions and various lost rates of inter coded macroblocks, the proposed method constantly outperforms the traditional integer-pixel boundary-matching based algorithm and demonstrates fairly good robustness, with a 1.0～2.5dB improvement in luminance PSNR compared to the traditional algorithm.