波变换

开发平台：
Matlab

denoise0615.m：源码内容
							% 所有实验结果汇总
I = imread('lena.png');                                             % 读入图像
X = im2double(I);                                                   % 转为双精度类型
% x = noise(X, 'gaussian', 0.005);                                    % 添加噪声 高斯噪声 方差 0.005
x = noise(X, 'gaussian', 0.01);                                     % 添加噪声 高斯噪声 方差 0.01
% x = noise(X, 'guassian', 0.02);                                     % 高斯噪声 方差 0.02
% x = noise(X, 'salt & pepper', 0.02);                                % 椒盐噪声 密度 0.02
% x = noise(X, 'salt & pepper', 0.05);                                % 椒盐噪声 密度 0.05
[C, S]      = wavedec2(x, 2, 'sym4');
thr         = Donoho(x);                                            % 计算Donoho全局阈值
thr_lvd     = Birge_Massart(C, S);                                  % 计算Bige-Massa策略阈值
alpha       = 0.5;
% 去噪 2层分解 sym4 小波--------------------------------------------------------
x_soft      = wdenoise(x, 'gbl', 's', thr, 'sym4', 2);              % 软阈值方法 Donoho 全局阈值
x_hard      = wdenoise(x, 'gbl', 'h', thr, 'sym4', 2);              % 硬阈值方法 Donoho 全局阈值
x_soft_lvd  = wdenoise(x, 'lvd', 's', thr_lvd, 'sym4', 2);          % 软阈值方法 Birge-Massart策略 计算的阈值 
x_hard_lvd  = wdenoise(x, 'lvd', 'h', thr_lvd, 'sym4', 2);          % 硬阈值方法 Birge-Massart策略 计算的阈值 
x1          = den1(x, 'sym4', 2, thr);                              % 半软阈值 原始方法
x1_5        = den1_5_1(x, 'sym4', 2, thr, 0.5*thr);                 % 半软阈值 + 均值滤波 
x1_9        = den1_9(x, 'sym4', 2, thr, 0.5*thr);                   % 半软阈值 + 指数衰减
x1_10       = den1_10(x, 'sym4', 2, thr, 0.5*thr);                  % 半软阈值 + 再次进行小波阈值去噪
x2          = den2(x, 'sym4', 2, thr, 0.4);                         % 软硬阈值的改进方法
x3          = den3(x, 'sym4', 2, thr, 5);                           % 广义小波阈值函数
x4          = den4(x, 'sym4', 2);                                   % 自适应特征阈值
% 计算峰值信噪比--------------------------------------------------------------
psnr_soft = PSNR(x_soft, X)                                         % Donoho公式 软阈值
psnr_hard = PSNR(x_hard, X)                                         % Donoho公式 硬阈值
psnr_soft_lvd = PSNR(x_soft_lvd, X)                                 % Birge-Massart策略 软阈值
psnr_hard_lvd = PSNR(x_hard_lvd, X)                                 % Birge-Massart策略 硬阈值
psnr1 = PSNR(x1, X)                                                 % 半软阈值
psnr1_5 = PSNR(x1_5, X)                                             % 半软阈值 + 均值滤波
psnr1_9 = PSNR(x1_9, X)                                             % 半软阈值  指数衰减
psnr1_10 = PSNR(x1_10, X)                                           % 半软阈值 再次进行小波阈值去噪
psnr2 = PSNR(x2, X)                                                 % 软硬阈值的改进方法
psnr3 = PSNR(x3, X)                                                 % 广义小波阈值函数
psnr4 = PSNR(x4, X)                                                 % 自适应特征阈值
% 显示图像---------------------------------------------------------------------
figure; imshow(x_soft);     title('软阈值方法 Donoho 全局阈值');
figure; imshow(x_hard);     title('硬阈值方法 Donoho 全局阈值');
figure; imshow(x_soft_lvd); title('软阈值方法 Birge-Massart策略');
figure; imshow(x_hard_lvd); title('硬阈值方法 Birge-Massart策略');
figure; imshow(x1);         title('半软阈值方法');
figure; imshow(x1_5);       title('半软阈值方法 线性衰减');
figure; imshow(x1_9);       title('半软阈值方法 指数衰减');
figure; imshow(x1_10);      title('半软阈值方法 再次进行小波去噪');
figure; imshow(x2);         title('软硬阈值函数的改进方法');
figure; imshow(x3);         title('广义小波阈值函数');
figure; imshow(x4);         title('自适应特征阈值');
% 非小波方法去噪------------------------------------------------------------------
f = ones(3);
x_mean1 = meanfilter(x,f);                                          % 均值滤波
x_med1  = medfilter(x, f);                                          % 中值滤波
% 计算峰值信噪比
psnr_mean1 = PSNR(x_mean1, X)                   
psnr_med1 = PSNR(x_med1, X)
figure; imshow(x_mean1);title('均值滤波1');
figure; imshow(x_med1); title('中值滤波1');

						