基于H.264的运动矢量插值修复算法研究

Motion vector interpolating recovery algorithm based on H.264

在视频编码新标准H.264中,一个16×16的帧间编码宏块可采用多种宏块分割模式来进行运动估计,其运动矢量场也更为精细。根据其运动矢量的邻域相关性提出了一种有效的运动矢量修复算法,可在压缩视频流在Internet上进行传输时发生丢包后,通过受损宏块时空域相邻宏块相关信息对丢失的运动矢量进行修复,并针对连续图像片丢失或整帧丢失情况进行处理。仿真实验表明,该方法可有效地抑制视频差错扩散,改善视频质量。

In new coding standard H.264, a 16× 16 inter macroblock can be divided into different block shapes for motion estimation, and motion vector filed becomes more elaborate. A motion vector interpolating recovery algorithm was proposed by the correlation between neighboring motion vectors. Since the bitstream of compressed video is often corrupted by channel errors of losing packet in Intemet, the lost motion vectors of damaged MB can be recovered. The algorithm also deals with the condition that current entire frame or plenty of consecutive slice have lost. Simulation result showed that the proposed algorithm could efficiently prevent error propagation and improve video quality.