Joint channel and source decoding for vector quantized images using turbo codes

This paper presents a novel joint decoding scheme for image transmission over noisy channels. The proposed decoding scheme includes two features of interaction between a turbo channel decoder and a vector quantization source decoder. First, the source decoder makes use of the channel soft outputs for better reconstruction; second, a feedback algorithm is designed for using source information to improve the error correction capability of the turbo decoder. Significant improvement in terms of the error correction capability and the quality of reconstructed images is achieved compared to a separately operating decoding system.