分级视频码流的三维精细粒度排序方法

为了降低网络延时、带宽变化以及信道丢包对视频传输的影响，提高视频重建质量，提出了一种基于三维精细粒度的可分级视频码流的排序方法．该方法以图像组为单位，通过计算空域、时域和信噪比域分层码流的失真，并依次选择失真较小的分层码流来完成对可分级视频码流的排序，保证接收端能够从收到的一定数量的数据包中恢复出最佳质量的图像．采用细化分层和改进迭代的方法，克服了已有算法当分层码流对应码率变化较大时失真计算不精确的问题，进而能够提供更加精细的分层码流排序特性．仿真结果表明，同已有方法相比，该方法能够有效地提高可分级码流的重建图像质量．

Improved 3D fine granularity priority ordering method for the scalable video bit stream

In order to reduce the effects of the network delay,bandwidth fluctuation and the packet loss and improve the quality of the reconstructed video,an improved 3D fine granularity priority ordering method for the scalable video bit stream is proposed.The method arranges the SVC coding layers according to their rate-distortion(RD) characteristics from spatial,temporal and SNR aspects within a group of picture(GOP),so that the optimal reconstructed video quality can be obtained when a certain number of packets are correctly received at the decoder.By layer refinement and improved iteration,the method overcomes the imprecision of the distortion calculation in the previous method when the variation of the layer bit stream is large,and provides finer granularity ordering characteristics for the layered bit stream.Simulation results show that the proposed method can effectively improve the reconstructed quality of SVC bit streams compared with existing methods.