论文部分内容阅读
目的分布式视频编码较其传统视频编码具有编码简单、误码鲁棒性高等特点,可以很好地满足如无人机航拍、无线监控等新型视频业务的需求。在分布式视频编码中,视频图像被交替分为关键帧和Wyner-Ziv帧,由于受到信道衰落和干扰等因素的影响,采用传统帧内编码方式的关键帧的误码鲁棒性远不如基于信道编码的WynerZiv帧。关键帧能否正确传输和解码对于Wyner-Ziv帧能否正确解码起着决定性的作用,进而影响着整个系统的压缩效率和率失真性能。为此针对关键帧在异构网络中的鲁棒性传输问题,提出一种基于小波域的关键帧质量可分级保护传输方案。方法在编码端对关键帧同时进行传统的帧内视频编码和基于小波域的Wyner-Ziv编码,解码端将经过错误隐藏后的误码关键帧作为基本层,Wyner-Ziv编码产生的校验信息码流作为增强层。为了提高系统的分层特性以便使系统的码率适应不同的网络条件,进一步将小波分解后图像的各个不同层的低频带和高频带组合成不同的增强层,根据不同信道环境,传输不同层的Wyner-Ziv校验数据。同时对误码情况下关键帧的虚拟噪声模型进行了改进,利用第1个增强层已解码重建的频带与其对应边信息来获得第2个和第3个增强层对应频带的更加符合实际的虚拟信道模型的估计。结果针对不同的视频序列在关键帧误码率为1%20%时,相比较于传统的帧内错误隐藏算法,所提方案可以提高视频重建图像的主观质量和整体系统的率失真性能。例如在关键帧误码率为5%时,通过传输第1个增强层,不同的视频序列峰值信噪比(PSNR)提升可达2 5 d B左右;如果继续传输第2个增强层的校验信息,视频图像的PSNR也可以提升0.5 1.6 d B左右;如果3个增强层的校验信息都传输的话,基本上可以达到无误码情况下关键帧的PSNR。结论本文所提方案可以很好地解决分布式视频编码系统中的关键帧在实际信道传输过程中可能出现的误码问题,同时采用的分层传输方案可以适应不同网络的信道情况。
Purpose Distributed video coding has the advantages of simple coding and high bit error rate robustness compared with traditional video coding, which can meet the needs of new video services such as UAV aerial photography and wireless monitoring. In distributed video coding, the video images are alternately divided into keyframes and Wyner-Ziv frames. Because of the influence of channel fading and interference, the error robustness of the keyframes adopting the traditional intraframe coding method is far less than that based on the Channel-coded WynerZiv frames. Whether the key frames can be correctly transmitted and decoded plays a decisive role in whether the Wyner-Ziv frames can be correctly decoded, thereby affecting the compression efficiency and rate-distortion performance of the entire system. Therefore, aiming at the robust transmission of key frames in heterogeneous networks, a key frame quality scalable protection and transmission scheme based on wavelet domain is proposed. The method carries on the traditional intra-frame video coding and the Wyner-Ziv coding based on the wavelet domain at the coding end. The decoding end uses the error-concealed key frame as the base layer and the Wyner-Ziv coding to generate the check information Code stream as an enhancement layer. In order to improve the system’s hierarchical characteristics in order to adapt the system’s code rate to different network conditions, the low-frequency and high-frequency bands at different layers of the wavelet decomposition image are further combined into different enhancement layers, which are transmitted according to different channel environments Wyner-Ziv calibration data for layers. At the same time, the virtual noise model of the key frame in the case of bit error is improved. By using the decoded and reconstructed frequency band of the first enhancement layer and its corresponding edge information, the virtual noise model of the second and the third enhancement layer is obtained more realistic Channel model estimation. Compared with the traditional intra-frame error concealment algorithm, the proposed scheme can improve the subjective quality of the reconstructed image and the rate-distortion performance of the whole system when the bit error rate of the video frame is 1% to 20% for different video sequences. For example, when the key frame error rate is 5%, the PSNR of different video sequences can be increased up to 25 dB by transmitting the first enhancement layer. If the second enhancement layer is continued The PSNR of the video and video images can also be increased by about 0.5 1.6 dB. If all the verification information of the three enhancement layers is transmitted, the PSNR of the key frame can be basically reached without error. Conclusion The scheme proposed in this paper can well solve the problem of bit errors that may occur in the actual channel transmission of key frames in distributed video coding systems. Meanwhile, the layered transmission scheme can adapt to the channel conditions of different networks.