MAP,MLE.rar_MAP去噪_ML_MLE去噪_map_图像噪声估计

clear clc %输入参数变量的值 p = 0.1; %序列{+1,-1}中+1出现的概率为p，－1出现的概率为1-p signal_power = 1; %信号功率为1 noise_power = 0.0; %噪声功率为0 noise_power_times = 50; %信噪比的次数 test_times = 100; %重复实验的次数 length = 1000; %序列的长度 SNR_dB = zeros(noise_power_times,1); %保留每次实验的SNR BER_mean_MAP = zeros(noise_power_times,1); %保留每次MAP实验BER的均值 BER_mean_ML = zeros(noise_power_times,1); %保留每次ML实验BER的均值 BER_all_MAP = zeros(test_times,1); %保留每次MAP实验的BER BER_all_ML = zeros(test_times,1); %保留每次ML实验的BER %运算过程 for i = 1: noise_power_times %信噪比的个数 noise_power = noise_power+i/10; %噪声功率N noise_mean = 0; %高斯白噪声的均值 noise_var = noise_power*0.5; %高斯白噪声的方差 for j = 1:test_times %进行重复实验 %信号序列 signal = randsrc(1,length,[-1 1;1-p p]); %randsrc()产生-1和1的贝努利随机序列，概率均为 0.5 %噪声序列 noise = normrnd(noise_mean,noise_var,[1,length]); %normrnd()产生N(noise_mean,noise_var)的高斯白噪声 %加噪声的信号序列 mixed_signal = signal + noise; judged_signal_MAP = zeros(size(signal)); %zeros()初始化序列的值为0，size()获得矩阵的行数和列数 judged_signal_ML = zeros(size(signal)); %zeros()初始化序列的值为0，size()获得矩阵的行数和列数 for k = 1:length %最佳MAP判决的信号序列 if mixed_signal(k) >= 0.25*noise_power*log((1-p)/p) %判决门限 judged_signal_MAP(k) = 1; else judged_signal_MAP(k) = -1; end %最佳ML判决的信号序列 if mixed_signal(k) >= 0 %判决门限 judged_signal_ML(k) = 1; else judged_signal_ML(k) = -1; end end %计算两个序列的差别，判决的信号和原来的信号相同的位置为 0 difference_MAP = signal - judged_signal_MAP; difference_ML = signal - judged_signal_ML; %计算错误bit的个数 bit_error_MAP = nnz(difference_MAP); %nnz()计算非0值的个数 bit_error_ML = nnz(difference_ML); %计算误比特率 BER_all_MAP(j) = bit_error_MAP / length; BER_all_ML(j) = bit_error_ML / length; end %对应for i = 1:times %计算误比特率的数学期望 BER_mean_MAP(i) = mean(BER_all_MAP); %mean()计算平均值 BER_mean_ML(i) = mean(BER_all_ML); %mean()计算平均值 SNR_dB(i) = 10*LOG10(signal_power/noise_power); end %绘出图形 figure(1); %figure()新建一个绘图窗口 plot(SNR_dB,BER_mean_MAP,'b-o'); %蓝线表示最佳MAP准则信噪比与无码率的曲线 hold on; plot(SNR_dB,BER_mean_ML,'r-*'); %红线表示最佳MAP准则信噪比与无码率的曲线