基于AWGN信道下Turbo码在静止图像传输系统中的应用

Application of Turbo Codes in Still Image Transmission Systems over AWGN Channel

数字图像经压缩后 ,在噪声信道中传输时很容易受到干扰 ,需要采用有较强纠错能力的信道编码来提高可靠性 .给出高斯白噪声信道 (AWGN)下静止图像传输系统结构 ,其中信源编码采用矢量量化编码 ,利用LBG算法生成码本 ,使用分裂法生成初始码本 ;信道编码采用Turbo码 .简要介绍由两个并行级联的递归系统卷积编码器和一个交织器组成的Turbo码编码器 ,由两个软输入软输出译码器串行级联为主体的Turbo码译码器的结构及原理 ;并详细介绍用于Turbo码译码的Log_MAP算法 .应用MATLAB仿真软件 ,给出 3种情况下的仿真结果 :无信道编码 ;信道编码采用卷积码 (2 ,1,6 ) ;信道编码采用生成多项式为 g =(7,5 ) 8,码率 =1/ 2 ,交织长度为 6 36 30 ,软输入软输出算法采用Log_MAP算法的Turbo码 .仿真结果表明 ,当Turbo码应用于有噪声的静止图像传输系统时 ,其性能优于卷积码 ,不仅提高整个系统的可靠性 ,还节省了系统发射功率 .

Compressed image bit streams are very sensitive to bit errors during transmission over a noise channel,which can severely degrade the quality of the image at the receiver.Therefore,the simplest solution to protecting the image information is to use powerful error-correction coding(channel coding),in order to improve the robustness against transmission errors over a noise channel.In this paper,the transmission system structure of the still image over an additive white Gaussian noise(AWGN)channel is given.Source coding uses Vector Quantization(VQ),LBG-VQ algorithm is used to create the codebook and requires an initial codebook,which is obtained by the splitting method;channel coding uses Turbo codes.Turbo codes were presented by C.Berrou in the International Conference on Communications in1993.Turbo codes achieve a bit error rate(BER)of10 -5 at a signal-to-noise ratio(SNR)of E b /N 0 ≥0.7dB,rate=0.5,length N=65536,iteration times are18,which is only0.7dB from the Shannon limit(limit is0dB when rate=0.5).In other words,Turbo coding is a novel form of channel coding capable of achieving a performance near the Shannon limit.The structure and principle of Turbo encoder and Turbo decoder are introduced briefly.The encoder of a Turbo code consists of two recursive systematic convolutional encoders jointed together by a random interleaver.The decoder is made up of two elementary decoders in a serial concatenation scheme,which adopts soft input/soft output(SISO)iterated decoding.Log-MAP algorithm using Turbo decoding is introduced in detail.For comparison,computer simulation experiments are done by using MATLAB for still image transmission systems over additive white Gaussian noise channels in three situations:(1)channel coding is not used;(2)the convolution code(2,1,6)is used as an error-correction code;(3)Turbo code is used as an error-correction code with the generator polynomial of a g=(7,5) 8 、rate of0.5、interleaver length N=65536、soft input/soft output(SISO)iterated decoding used Log-MAP Algorithm.Simulation results show that the performance of Turbo codes used in still image transmission systems outperforms that of the convolutional codes.Turbo codes obtain a significant coding gain,which not only improves the reliability of communication systems,but also saves power budget.