Journal of Electronic ImagingProgressive image transmission with genetic algorithm–based optimization and multiple cyclic redundancy check structure
|Format||Member Price||Non-Member Price|
We consider the transmission of progressive image data over noisy channels when the coded packet size is fixed. The concatenated cyclic redundancy check (CRC) codes and ratecompatible punctured turbo codes are used for error control and detection. In such an application, the distortion-based optimal channel rate allocation for unequal error protection is complex. We first propose a suboptimal genetic algorithm–based method that not only largely reduces the optimization complexity but also obtains performance approaching to the results of a brute force search. In addition, because a large packet size is usually applied when turbo codes are used due to the fact that the coding gain is proportional to the packet size for a given code rate, a single remaining bit error after channel decoding may result in CRC failure and hence the discard of the entire packet. Therefore, we further propose a multiple-CRC structure for certain data packets so that more correctly decoded data could be used in source decoding. The promising performance of the proposed scheme has been demonstrated through simulation.