Empirical channel matched quantizer design and UEP for robust image transmission

Summary form only given. Channel matched quantization for image transmission over time varying channels reduces the effects of channel errors. The presence of variable length codes in compression standards like the JPEG cause error propagation due to bit errors. Unequal error protection (UEP) schemes have emerged as an effective method to combat catastrophic loss in the received signal due to burst and random errors. An empirical channel matched quantizer design algorithm that jointly optimizes the distortion due to quantization-channel noise and a new efficient UEP scheme for image transmission are proposed. The baseline JPEG encoder is used to compress the 8-bit gray level images before transmission. A slow frequency non-selective Rayleigh fading channel is considered in this study. A quantization table optimized for human visual quality is used for very low channel bit error rates. For higher bit error rates, the quantization table is matched to the channel conditions by multiplying its entries by the optimal quantization multiplication factor, M^* such that the average number of received image blocks in error is minimized. M^* is computed for each bit error rate ranging from 10^-4 to 10^-1 through empirical modeling of the trade-off between the quantization and the channel noise. In order to enhance the performance of the proposed system, a new UEP scheme that limits the error propagation due to variable length encoding is used. This scheme works by packing the output bits of the JPEG coder into slots of fixed size