A study of Systematic Lossy Error Protection Architecture in H.264 Video Transmission

Abstract

There are inevitable packet losses when the video transmits over the network, and that leads to the video decoder can’t decode correctly, hence results in unpleasant video quality. The new generation of H.264 video compression standard exhibits high compression efficiency, but occasionally incurs more serious error propagation when an error occurs and this makes the video quality degrade drastically. Therefore, the development of an effective error protection mechanism for the video transmission is very important.

Systematic Lossy Error Protection (SLEP) is a robust error resilient mechanism which uses Wyner-Ziv coding to protect the video bitstream. SLEP avoids the “cliff” effect, i.e., the rapid degradation of video quality when too many packets are lost and the FEC code fails to recover them, from which the FEC system suffers and produces graceful degradation of the decoded video quality with the increase of error probability. In this thesis, we study and implement the SLEP architecture on H.264 and try to get the best trade-off between the error resilience and decoded video quality. We propose an adaptive SLEP structure for further improvement of error resilience capability by generating redundant slices with different coding methods for different channel conditions.

As expected, the SLEP architecture achieves graceful degradation of video quality and provides acceptable visual quality in the presence of increasing network packet losses.