中文摘要：

为解决多视点视频压缩（MVc）导致视频整帧丢失风险增大的问题，基于MVc的特点提出了一种4方向和多参考帧的错误掩盖算法。首先，在同一视点内，将相邻帧沿其各自运动矢量外推，得到视点内的最佳运动矢量；在视点间采用同样的方法，得到最佳视差；以外推的重叠面积为判断依据，在视点内错误掩盖和视点间错误掩盖之间选择较优的掩盖方式；最后，在MVC已有块模式的基础上引入新的块组合模式，合并具有相同错误掩盖方式、运动矢量的绝对差值小于阈值的相邻块，以减小图像的块效应。实验结果表明，与已有算法相比，本文算法有效利用了MVC的内部机制，改善了重建帧的客观指标，增强了主观质量。

英文摘要：

Multiview video compress （MVC） provides high compression efficiency for the multi-view vide- o coding,and as a result,suffers higher possibility of entire frame loss. To solve this problem,this paper proposes a novel error concealment algorithm based on four reference frames. Firstly, within the lost framels view,adjacent frames are extrapolated to the lost frame along their motion vectors to get the op- timal motion vectors of the lost frame. And then,optimal disparity can also be obtained by the same ap- proach. An compromise based on the size of overlapping area has been proposed to choose the better er- ror concealment mode between intraview error concealment and interview error concealment. Finally, this paper introduces new block modes for the corrupt frame except for traditional size 4 × 4 to combine smaller blocks with the same error concealment mode and similar motion vectors（disparities）,and ulti- mately,mitigates blocking artifacts. Simulation results illustrate that the proposed method can gain a peak signal-to-noise ratio （PSNR） of 2.73 dB higher than the conventional motion copy method on aver- age, and of up to 1.92 dB higher than conventional motion vector extrapolation method. Compared with existing methods with the same computation complexity, the proposed algorithm takes advantage of the mechanics of MVC efficiently and performs ameliorative objective indicators and has better subjective quality for the concealed frame as well.