An integrated joint source-channel decoder for MPEG-4 coded video

Qingyu Chen, K. P. Subbalakshmi

Research output: Contribution to journalConference articlepeer-review

10 Scopus citations

Abstract

The MPEG-4 video coding standard uses variable length codes (VLCs) to encode the indices of intra and inter macroblocks after discrete cosine transform and quantization. Although VLCs can achieve good compression, they are very sensitive to channel errors. Joint source-channel decoding (JSCD) is emerging as an efficient method for dealing with this sensitivity to channel errors. This paper proposes an integrated joint source-channel decoder (I-JSCD) for first order Markov sources coded with Huffman codes and convolutional codes over a binary symmetric channel. The proposed decoder combines the source state space of the Huffman decoder and channel state space of the Viterbi decoder together to construct a joint decoder with a Viterbi-like structure. We applied this I-JSCD to VLCs of both inter and intra macroblocks in MPEG-4 coded video. Experiments indicate that the proposed decoder gives significant improvements (maximum 7 dB) than a separate scheme, where a constrained joint source-channel decoder is concatenated with a Viterbi decoder at the same rate.

Original languageEnglish
Pages (from-to)347-351
Number of pages5
JournalIEEE Vehicular Technology Conference
Volume58
Issue number1
StatePublished - 2003
Event2003 IEEE 58th Vehicular Technology Conference, VTC2003-Fall - Orlando, FL, United States
Duration: 6 Oct 20039 Oct 2003

Fingerprint

Dive into the research topics of 'An integrated joint source-channel decoder for MPEG-4 coded video'. Together they form a unique fingerprint.

Cite this