Iterative approximation on source and channel coding for images transmission over noisy channels

In this paper we present an application of iterative approximation on source and channel coding for images transmission over noisy channels. The proposed approach to this joint source-channel coding includes a new PSNR evaluator. With this evaluator, an early termination technique is used to reduce the iterative decoding loops. In additions, we propose two schemes for different compressed image. One is for the low bitrate image quality and the other is for the high bit-rate image quality. For each time of iteration, it gains the average PSNR up to 8/spl sim/9 dB in the same bit-rate with the vary channel condition. In order to achieve the corresponding performance, the optimal computation on iterations is investigated in these two schemes. With the lower quality image transmission, there is little PSNR improvement on more than 5 iterations. With the higher quality image transmission, a less-iteration decoding is applied with the early termination supporting. The results show our approach can decrease the complexity of the source decoder. This technique is suitable for real time system.