Combined source-channel image coding using tree-structured vector quantization and sequential decoding

The authors propose a flexible system for combined source-channel image coding for the AWGN (additive white Gaussian noise) channel. The system can operate over a wide range of channel SNR (signal-to-noise ratio) values by changing the way it allocates bits between source and channel coding. The system is based on tree-structured vector quantization and soft-decision sequential decoding with a maximum a posteriori metric. The rate at which images are communicated and the channel bit rate are constant in all modes of operation. The simulated performance of the system is discussed, and the metric it employs for sequential decoding is derived.< >