中文摘要：

针对网络传输中整帧丢失的情况，提出了一组基于编解码端联合控制的错误隐藏机制．在解码端，通过运动连续性假设，从前一帧恢复出相应的像素值，针对剩余的无效像素点，将整帧以4×4块进行划分并且分类，使用块预测匹配以及帧内插值，恢复它们的像素值．在编码端，采用长期参考帧和短期参考帧相结合的方法，通过信道的反馈信息，对参考帧进行调整，抑制错误的扩散．实验证明，该方法不论在主观视觉还是客观峰值信噪比（PSNR）上都取得了很好的效果，而且有效地抑制了错误的扩散；同时计算复杂度不高，能很好地满足实时性要求．

英文摘要：

An error concealment mechanism based on encoder and decoder was proposed for whole frame packet lost in network transmission. In decoder, some pixels were directly recovered from the previous frame according to the hypothesis of motion continuity. Each 4× 4 block was classified to different types in order to recover other invalidated pixels. Then two different methods,one based on motion prediction and block match, another based on spatial interpolation, were used to recover the invalidated pixels. In encoder, both long reference frames and short reference frames were adopted. The reference frames were adjusted to stop error propagation by using the feedback of the receiver. Experimental results show that the method outperforms on both visual quality and peak signal-to-noise ratio （PSNR）, and can be effective to stop error propagation. Meanwhile the method can satisfy the real time request.