联合编码图像传输码率分配研究

Rate Allocation Algorithm for Image Transmission Based on Joint Source and Channel Coding

嵌入式压缩编码算法因易受误码扩散影响而无法用于图像的可靠传输,信源信道联合编码已被证明是解决此问题的有效方法。针对联合编码中的关键问题-码率分配,提出了一种基于信源码率最优的快速算法,首先求解对应的等差错保护(EEP)问题,并用该解初始化不等差错保护(UEP)问题,然后采用前后向码率搜索得到最佳码率分配。该方案计算复杂度低,重建图像质量高。仿真结果表明,接收端重建图像的PSNR值始终在34dB以上,同时波动范围小于5dB。

Transmission of images compressed by embedded source coders is difficult in a noisy channel, mainly due to the propagation of errors during decoding.Excellent results for this problem are made possible through joint source and channel coding （JSCC）.A fast computation algorithm is proposed for the key component of JSCC： rate allocation.Initially, the EEP （Equal Error Protection） problem is solved, and then protection is iteratively decreased at the beginning of the bitstream and increased from the end. In that way, protection schemes given with consecutive iterations converge to the optimal UEP（Unequal Error Protection）.Simulation results show that while keeping the computation cost low, the proposed UEP scheme introduces improvement over EEP and Fixed-UEP.The PSNR（Peak Signal to Noise Ratio）of reconstructed images is always above 34riB and the fluctuation is less than 5dB.