基于感兴趣区域的自适应帧内更新编码算法

针对视频压缩码流对信道差错异常敏感的问题，提出了一种基于感兴趣区域的自适应帧内更新编码算法。该算法利用人眼的视觉感知特性设计感兴趣区域提取模型，根据信道差错累积情况自适应调整帧内更新编码策略，将更少的失真分配给人眼感兴趣区域，提高差错信道下传输视频图像的主客观质量。实验结果表明，与基于端到端失真的帧内更新算法相比，在不同的信道丢包条件下，所提算法可以获得更好的主客观质量，感兴趣区域的峰值信噪比（PSNR）平均提高0.87 dB左右，提高了差错信道下视频通信的鲁棒性。

Adaptive intra update coding algorithm based on region of interest video coding

In view of the sensitivity of video coding stream to channel error,an intra refresh algorithm based on region of interest video coding is proposed. The algorithm first estimates the model for extracting region of interest using human visual perception,and then it chooses intra coding strategy adaptively according to the accumulation of errors for every macroblock,allocating less distortion to region of interest to improve the subjective and objective visual quality under lossy channel. Simulation shows that under various packet loss rates,the proposed algorithm can obtain a better subjective and objective quality of the reconstructed video and outperforms 0.87 dB in region of interest than the end-to-end intra update algorithm,thus improving the robustness of video communications under lossy channel.