A Distributed Video Coding approach for Multiple Description video transmission over lossy channels

Recent works have proposed the combination of Multiple Description Coding (MDC) schemes with the novel Distributed Video Coding (DVC) paradigmin order to enable robust video transmission over lossy channels. DVC schemes permit mitigating the drifting of channel distortion that characterizes the traditional MDC schemes and prove to be extremely effective at high packet loss rates. The paper presents a Multiple Description Distributed Video Coder (MDDVC) that combines a polyphase-subsampling MDC approach with a novel hybrid Distributed Source Coding technique that processes the video signal both in the pixel and in the transform domain. Tests on simulated lossy channels show that at high loss probabilities the PSNR value of the reconstructed video sequence improves with respect to its traditional MDC counterpart. Moreover, an adaptive approach is also presented combining both the traditional and the DSC-based MDC schemes in order to maximize the quality of the sequence reconstructed at the decoder for all the network conditions.