基于残差恢复的错误隐藏算法

A Novel Error Concealment Algorithm Based on Residue Restoration

提出一种基于残差恢复的错误隐藏算法。在运动补偿（MC）错误隐藏算法（MCECA）的基础上，利用错误宏块周围已正确接收和解码的信息，估计宏块的残差数据，并将其加到MC恢复的宏块上，实现更为准确的错误隐藏，以提高图像的主客观质量。实验结果表明，该算法与MCECA相比较，恢复图像的客观指标峰值信噪比（PSNR）将能提高1～4dB，而主观质量提高更明显。

A novel error concealment algorithm based on residue restoration is proposed, which utilizes the boundary information to estimate the residue of macro-block（MB）. In the algorithm,we firstly use the motion compensation error concealment algorithm（MOEOA） to restore error MB. In order to reconstruct error MB more accurately,we use the boundary pixels to construct the residue data,then employ the 5-tap FIR filter to estimate MB residue,and at last add the estimated residue to the restoration MB by temporal replacement. The experimental results show that the proposed algorithm can efficiently improve the video equality （by 1-4 dB） compared with MCECA.