| As the continuous development of streaming media technology and network transmissiontechnology, as well as user demand rises, digital video communications are widely utilized invarious fields. In order to balance the contradiction between storage capacity and transmissionbandwidth, the original video possessed large number of data must be compressed effectivelybefore transmission. As higher compression efficiency and better network-friendly, H.264/AVC iswidely used in video communications. Although many new technologies adopted in H.264/AVCare used to improve coding efficiency, they leads to compressed video signals are very sensitive totransmission errors. Even a bit error may lead to the loss of synchronization and decoding errorsdiffusion, and it will causes the bit streams cannot be decoded properly or cannot be decoded. Itwill result in a sharp decline in the quality of reconstructed video. Therefore, it is important toimprove error recovery and anti-interference ability to error in video communication.The paper mainly focuses on the error resilience algorithm for H.264/AVC. The importantfeature of the video data is embedded in bit stream using the reversible information hidingtechnique, and it is utilized for error recovery at the decoder side. The error resilience methods ofour research is illustrated as follow.Firstly, the paper analyzes and discusses the video encoding technology, as well as the causeof the bit error and packet loss. And then some classical error resilience algorithms based on H.264are analyzed and researched deeply.Secondly, by in-depth studing for block feature information used in the error blockreconstruction and error recovery mechanism in video communication, an video error resiliencealgorithm based on generalized difference reversible information hiding is proposed. The algorithmtakes motion vector (MV) of each macroblock (MB) in Intra-Frame as the important information,and it is embedded into the coefficients of quantized discrete cosine transform by a reversibleinformation hiding technique. The feature extracted at the decoded side is utilized error Resilience.Simulation results show that, compared to similar methods, on the one hand, the proposedmethod increases the visual quality of stego-video. At the different of QP, the proposed algorithmincreases average4.03dB~5.34dB than the error recovery algorithm based on histogram shiftingdata hidden algorithm (RDHEC) between the viewpoints in the same parameter. On the other hand,it effectively improve the error recovery quality of the I-frame. Compared with RHDEC, in thecase of Intra-frame MBs in the loss rate of20%, the proposed algorithm improves the errorrecovery quality average about0.075dB~0.46dB. While in the loss of20%, it can increases thevisual quality average about1.00dB~2.88dB. Thirdly, combined with new features of H.264, the paper proposed an error recoveryalgorithm base on intra prediction mode. The prediction modes of each MB in Intra-Frame as afeature are embedded in the coefficients of quantized discrete cosine transform. Feature data istransmitted to the decoder side in the service of error recovery. The simulation results on JM-12.0reference software shows that, the algorithm has achieved relatively good results for the presenceof a scene change or not. In the loss rate of10%and20%respectively, the proposed algorithmincreases the quality average on3.24dB and2.79dB respectively than weighted interpolationalgorithm (JM) between the viewpoints in the same parameter. While there is a scene change, theproposed methed improve increases the quality average on1dB compared with RDHECalgorithm.And from a subjective effect, this algorithm can guarantee the quality of reconstructionvideo decline in a graceful manner. |
PDF Full Text Request