In this work. we propose a Multiple Description (MD) ciding system for video streams. In particular. our scheme originates four descriptions from the spatially downsampled polyphase components ofthe original frames. Each description is compressed independently with the recent H264lAVC video coding standard. it i s packetized and sent over an error prone network. In case of errors i n one or more descriptions. appropriate concealing i s applied at the receiver, before insertion of the corrected Crames into the corresponding receiver frame buffers. We propose and compare different concealment solutions and a post processing stage to attenuate visual effects related to MD coding. We analyze the trade off between robustness to channel errors and coding efficiency. comparing the proposed technique with Single Description (SD) video coding with H264lAVC. Experimental results validate the effectiveness ofthe proposed scheme.
We consider the problem of robust video streaming over networks that support QoS differentiation, such as the 802.11e wireless network infrastructure. We consider the benefits obtained matching the H.264 Data Partitioning (DP) mode with such a QoS-based interface. We compare this solution with an innovative scheme which combines a Multiple Description (MD) coding framework with a QoS-based network. Results are reported using both a simple IID channel model and a more realistic wireless network, simulated using the OmNet++ network simulator.
Multiple Description (MD) coding has recently been recognized as an interesting option for data communication over lossy networks. Among the proposed techniques, Multiple Description using frames has received attention because of its possible advantages over error correcting codes. In frame-based MD, the input signal is analyzed with a redundant filterbank and synthesized with a dual filterhank which is the pseudo-inverse of the analysis linear operator. In this paper, we consider the problem of signal reconstruction in the presence of coefficient loss, which requires to compute the pseudo-inverse operator relative to the received coefficients after random losses. The computation of the dual operator using known techniques can be totally unpractical for many applications. We propose a novel algorithm forthe computation of the dual frame which can be implemented with reasonable complexity and allows for low-delay reconstruction ofthe input stream.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.