数字水印matlab程序.docx

clear all;close all;clc;V=double(imread(D:lenalena.jpg);imshow(mat2gray(V);i u=size(V); %计算V的规格r=100; %设置分解矩阵的秩W=rand(i,r) %初始化WH，为非负数H=rand(r,u)maviter=100; %最大迭代次数for iter=1:maviter W=W.*(V./(W*H)*H); %注意这里的三个公式和文中的是对应的 W=W./(ones(i,1)*sum(W); H=H.*(W*(V./(W*H);endimg_V=W*H;figure;imshow(mat2gray(img_V);首先读入原始图象并设置参数，然后嵌入水印信息，程序代码如下：clear%size=256;block=8;blockno=size/block;LENGTH=size*size/64;Alpha1=0.02;Alpha2=0.1; T1=3; I=zeros(size,size); D=zeros(size,size); BW=zeros(size,size); block_dct1=zeros(block,block);%产生水印序列并对其排序randn(seed,10);watermark1=randn(1,LENGTH);subplot(2,2,1);plot(watermark1);title(watermarc:Gaussian noise);subplot(2,2,3);title(edge of origine image)Y0,I0=sort(watermark1);%读入原图象trueImage=imread(H:Documents and SettingssunhwMy DocumentsMy Picturesbiaozhun.bmp);alfa=.1;LENGTH=2500;subplot(2,2,2);imshow(trueImage);title(origine image:I);%对原图象进行DCT变换dctF1=dct2(H:Documents and SettingssunhwMy DocumentsMy Picturesbiaozhun.bmp);m,n=size(dctF1);%找出水印嵌入位置（幅值较大的n个频域成分）A=dctF1(:);Y1,I1=sort(A);x=m*n;k=LENGTH;M=zeros(x,1);%修改幅值较大的n个频域成分的幅值，嵌入水印（因为两个问题不同，所以有两个注释符）for i=1:xif k=1M(x)=Y1(x)*(1+alfa*Y0(k);k=k-1;elseM(x)=Y1(x);endx=x-1;endN=zeros(x,1);x=m*n;for i=1:xN(I1(i)=M(i);enda=1;for j=1:nfor i=1:mdctF2(i,j)=N(a);a=a+1;endend%DCT反变换，得到嵌入水印的图象idctF1=idct2(dctF2);subplot(2,2,4);imshow(log(abs(idctF1), );title(embeded image:D);%endI=imread(D:lena1.jpg);disp(I);I=double(I)/255;disp(I);I=ceil(I);%显示水印图像%figure(1);subplot(2,3,1);imshow(I),title(水印图像)dimI=size(I);rm=dimI(1);cm=dimI(2);%5以下生成水印信息 %mark=I;alpha=0.05;V=imread(D:lenalena.jpg);i u=size(V); %计算V的规格r=100; %设置分解矩阵的秩W=rand(i,r) %初始化WH，为非负数H=rand(r,u)maviter=100; %最大迭代次数for iter=1:maviter W=W.*(V./(W*H)*H); %注意这里的三个公式和文中的是对应的 W=W./(ones(i,1)*sum(W); H=H.*(W*(V./(W*H);endk1=H;psnr_cover=double(V);subplot(2,3,2),imshow(a0,),title(载体图像);r,c=size(a0);cda0=blkproc(a0,8,8,dct2);% 嵌入 %cda1=cda0; % cda1 = 256_256for i=1:rm% i=1:32 for j=1:cm% j=1:32 x=(i-1)*10;y=(j-1)*10; if mark(i,j)=1 k=k1; else k=k2; end cda1(x+1,y+8)=cda0(x+1,y+8)*(1+alpha*k(1); cda1(x+2,y+7)=cda0(x+2,y+7)*(1+alpha*k(2); cda1(x+3,y+6)=cda0(x+3,y+6)*(1+alpha*k(3); cda1(x+4,y+5)=cda0(x+4,y+5)*(1+alpha*k(4); cda1(x+5,y+4)=cda0(x+5,y+4)*(1+alpha*k(5); cda1(x+6,y+3)=cda0(x+6,y+3)*(1+alpha*k(6); cda1(x+7,y+2)=cda0(x+7,y+2)*(1+alpha*k(7); cda1(x+8,y+1)=cda0(x+8,y+1)*(1+alpha*k(8); endend% 嵌入水印后图像 %a1=blkproc(cda1,8,8,idct2);a_1=uint8(a1);imwrite(a_1,withmark.bmp,bmp);subplot(2,3,3),imshow(a1,),title(嵌入水印后的图像);size=256;block=8;blockno=size/block;%一行有32格LENGTH=size*size/64;Alpha1=0.025;Alpha2=0.1;T1=3;I=zeros(size,size);%产生全矩阵D=zeros(size,size);BW=zeros(size,size);Block_dct1=zeros(block,block);%产生水印，并显示水印信息；subplot(3,2,1);Info=dcf;InfoStrSize=length(Info);%将字符串转换为位数组array=zeros(1,InfoStrSize*8);for m=1:InfoStrSizeInfochar=double(Info(m);% c为99for n=1:8 array(8*(m-1)+n)=bitget(Infochar,n);%获得Infochar第n位的值endendplot(array);title(原始水印信息);%显示原图subplot(3,2,2);i=imread(lena.bmp);imshow(i,);title(原始图像)%显示prewitt为算子的边缘图BW=edge(i,prewitt);%BW=edge(I,Roberts);%BW=edge(I,Sobel);%BW=edge(I,zerocross);subplot(3,2,3);imshow(BW);Title(原始图像边缘图);%嵌入水印l=1;k=1;for m=1:blocknofor n=1:blockno x=(m-1)*block+1; y=(n-1)*block+1;%算出每格图像的坐标（x,y）,block=8,8*8的图像小格 block_dct1=H(x:x+block-1,y:y+block-1);%取原始图像小格中的像素点到block_dct1矩阵中。block_dct1=dct2(block_dct1);%对二维数组进行离散余弦变换。dct是有损压缩如jpeg使用的技术。Dct是可逆的运算 BW_8_8=BW(x:x+block-1,y:y+block-1);%得到边界矩阵。 if m=1|nT1 Alpha=Alpha2; %block_dct1(1,1)=block_dct1(1,1)*(1+Alpha*mark(k); if l=(InfoStrSize*8) block_dct1(1,1)=block_dct1(1,1)*(1+Alpha*array(l); l=l+1; end else Alpha=Alpha1; % block_dct1(1,1)=block_dct1(1,1)*(1+Alpha*mark(k); end Block_dct=idct2(block_dct1); D(x:x+block-1,y:y+block-1)=Block_dct; k=k+1;endend%显示嵌入水印后的图像subplot(3,2,4);imshow(D,);title(嵌入水印图像)%保存该图像D=uint8(D);imwrite(D,marked.bmp);2：%提取水印 D=imread(marked.bmp); D=double(D);I=imread(lena.bmp);I=double(I);array2=zeros(1,InfoStrSize*8);K=1;l=1;for m=1:blocknofor n=1:blockno x=(m-1)*block+1; y=(n-1)*block+1;%算出每格图像的坐标（x,y）,block=8,8*8的图像小格 block_dct1=I(x:x+block-1,y:y+block-1);%取原始图像小格中的像素点到block_dct1矩阵中。 block_dct2=D(x:x+block-1,y:y+block-1);Block_dct1=dct2(block_dct1);%对二维数组进行离散余弦变换。dct是有损压缩如jpeg使用的技术。Dct是可逆的运算Block_dct2=dct2(block_dct2); BW_8_8=BW(x:x+block-1,y:y+block-1);%得到边界矩阵。 if m=1|nT1 Alpha=Alpha2; %block_dct1(1,1)=block_dct1(1,1)*(1+Alpha*mark(k); if lrow i1=mod(i1,row); end if j1colum j1=mod(j1,colum); end if i1= =0 i1=row; end if j1= =0 j1=colum; end w1(i1,j1)=w0(i,j); end end w0=w1;end r=w0;%提取水印for p=1:size/Bfor q=1:size/Bx=(p-1)*B+1; y=(q-1)*B+1;if (I_W(x,y)-P(x,y)0F(p,q)=1;elseF(p,q)=0;endendendfigure(5); imshow(F,); title(提取出的水印);%攻击实验disp(input you choice according to the following image processing operation:);disp(0-exit);disp(1-smoothing patterns);%添加噪音disp(2-adding uniorm noise 添加噪音); %滤波disp(3-adding filter 10 10 滤波);%剪切disp(4-cutting part of the image 剪切);%压缩disp(5-10 quality JPEG compressing 压缩); %旋转45度disp(6-rotate 45 旋转); %d=input(please input you choice(1,2,3,4,5,6):);while d=0switch d case 1watermark_detect(idctF1,Y1,I0,waterMark1);case 2 WImage2=idctF1; noise0=10*rand(size(WImage2); WImage2=WImage2+noise0; figure;imshow(WImage2, );title(adding uniform noise 添加噪音); watemark_detect(WImage2,Y1,I0,waterMark1);case 3 WImage3=idctF1; H=fspcial(gaussian高斯,10,10,5); WImage3=imfilter(WImage3,H); figure; imshow(WImage3, ); title(through filter 10,10 滤波); watemark_detect(WImage3,Y1,I0,waterMark1);case 4 WImage4=idctF1; WImage4(1:128,1;128)=256; figure; imshow(WImage4); title(cutting part of the image 剪切); watemark_detect(WImage4,Y1,I0,waterMark1); case 5 WImage5=idctF1; WImage5=im2double(WImage5); cnum=10; dctm=dctmtx(8); p1=dctm; p2=dctm.; imageDCT=blkproc(WImage5,8,8,p1*p2*x,dctm,dctm.); DCTvar=im2col(imageDCT,8,8,distinct).; n=size(DCTvar,1); DCTvar=(sum(DCTvar.*DCTvar)-(sum(DCTvar)/n).2)/n; dum,order=sort(DCTvar); cnum=64-cnum; mask=ones(8,8); mask(order(1:cnum)=zeros(1,cnum); im88=zeros(9,9); im88(1:8,1:8)=mask; im128128=kron(im88(1:8,1:8),ones(16); dctm=dctmtx(8); p1=dctm.; p2=mask(1;8,1:8); p3=dctm;Wimage5=bikproc(imageDCT,8,8,p1*(x.8p2)*p3,dctm.,mask(1:8,1:8),dctm); figure; imshow(Wimage5); title(JPEG Image 压缩); watemark_detect(WImage5,Y1,I0,waterMark1);case 6 WImage6=idctF1; WImage6=imrotate(WImage6,45,bilinear,corp); figure; imshow(Wimage6); title(rotate 45 旋转); watemark_detect(WImage6,Y1,I0,waterMark1);case 0 break;otherwise error(you have a valid value(您的输入错误);endd=input(please input you choice(请输入您的选择):);end%结束f=imread(watermark.jpg);%将含水印图像f归一化,以便于攻击处理。m=max(max(I_W);I_W=double(I_W )./double(m);%-攻击-attack=0;switch attack case 0, attackf= I_W; att=未攻击; case 1,%1. JPEG 压缩imwrite(I_W,attackf.jpg,jpg,quality,30);attackf=imread(attackf.jpg);attackf=double(attackf)/255;att=JPEG压缩; case 2,% %2. 高斯低通滤波h=fspecial(gaussian,3,1);attackf=filter2(h, I_W);att=高斯低通滤波; case 3,%3. 直方图均衡化attackf=histeq(I_W);att=直方图均衡化; case 4,%4. 图像增亮attackf=imadjust(I_W,0.4,1);att=图像增亮; case 5,%5. 图像变暗attackf=imadjust(I_W,0,0.85);att=图像变暗; case 6,%6. 增加对比度attackf=imadjust(I_W,0.3,0.6,);att=增加对比度; case 7,%7. 降低对比度attackf=imadjust(I_W,0.2,0.8);att=降低对比度; case 8,%8. 添加高斯噪声attackf=imnoise(I_W,gaussian,0,0.01);att=添加高斯噪声; case 9,%9. 椒盐噪声attackf=imnoise(I_W,salt & pepper,0.06);att=椒盐噪声; case 10,%10. 添加乘积性噪声attackf=imnoise(I_W,speckle,0.08);att=添加乘积性噪声;end;%-攻击后处理-I_W=attackf.*double(m);figure(2);imshow(uint8(I_W);%显示水印嵌入图攻击后效果title(att);imwrite(uint8(I_W),watermark.jpg);%-提取水印-% clear;a=imread(watermark.jpg);t=sdwt_ex(double(a),db2,tkey);%根据密钥树分解w,map=extract(t,tkey);%抽取水印r,c=size(w);figure(3);for i=1:r subplot(ceil(r/3),3,i) imshow(255-100*abs(uint8(reshape(w(i,:),map(1),map(2); title(strcat(抽取水印图,num2str(i);end;V=imread(D:lenalena.jpg);Alfa=0.05;LENGTH=2500; i u=size(V); %计算V的规格r=100; %设置分解矩阵的秩W=rand(i,r) %初始化WH，为非负数H=rand(r,u)maviter=100; %最大迭代次数for iter=1:maviter W=W.*(V./(W*H)*H); %注意这里的三个公式和文中的是对应的 W=W./(ones(i,1)*sum(W); H=H.*(W*(V./(W*H);endY1,10=sort(H)L=imread(D:lena1.jpg);Y0,10=sort(L)x=m*n;k=LENGTH;M=zeros(x,1);%修改幅值较大的n个频域成分的幅值，嵌入水印（因为两个问题不同，所以有两个注释符）for i=1:x if k=1 M(x)=Y1(x)*(1+alfa*Y0(k); k=k-1; else M(x)=Y1(x); end x=x-1;endN=zeros(x,1);x=m*n;for i=1:x N(I1(i)=M(i);enda=1;for j=1:nfor i=1:m N(i,j)=N(a); a=a+1;endend%重构图像，得到嵌入水印的图象B=W.*H;figure(1);imshow(B);title(嵌入水印后的图象);endclear all; clc; start_time=cputime; % 读取水印图像 % I=imread(mark.bmp); I=rgb2gray(I); I=double(I)/255; I=ceil(I); %显示水印图像% figure(1); subplot(2,3,1); imshow(I),title(水印图像) dimI=size(I); rm=dimI(1);cm=dimI(2); %5 以下生成水印信息 % mark=I; alpha=50, k1=randn(1,8); k2=randn(1,8); a0=imread(lena.bmp); psnr_cover=double(a0); subplot(2,3,2),imshow(a0,),title(载体图像); r,c=size(a0); cda0=blkproc(a0,8,8,dct2); % 嵌入 % cda1=cda0; % cda1 = 256_256 for i=1:rm % i=1:32 for j=1:cm % j=1:32 x=(i-1)*8;y=(j-1)*8; if mark(i,j)=1 k=k1; else k=k2; end cda1(x+1,y+8)=cda0(x+1,y+8)+alpha*k(1); cda1(x+2,y+7)=cda0(x+2,y+7)+alpha*k(2); cda1(x+3,y+6)=cda0(x+3,y+6)+alpha*k(3); cda1(x+4,y+5)=cda0(x+4,y+5)+alpha*k(4); cda1(x+5,y+4)=cda0(x+5,y+4)+alpha*k(5); cda1(x+6,y+3)=cda0(x+6,y+3)+alpha*k(6); cda1(x+7,y+2)=cda0(x+7,y+2)+alpha*k(7); cda1(x+8,y+1)=cda0(x+8,y+1)+alpha*k(8); end end % 嵌入水印后图像 % a1=blkproc(cda1,8,8,idct2); a_1=uint8(a1); imwrite(a_1,withmark.bmp,bmp); subplot(2,3,3),imshow(a1,),title(嵌入水印后的图像); disp(嵌入水印处理时间); embed_time=cputime-start_time, % disp(对嵌入水印的图像的攻击实验，请输入选择项：); disp(1-添加白噪声); disp(2-高斯低通滤波); disp(3-JPEG 压缩); disp(4-图像剪切); disp(5-旋转10度); disp(6-直接检测水印); disp(其他-不攻击); d=input(请输入选择（1-6）:); start_time=cputime; figure(1); switch d case 6 subplot(2,3,4); imshow(a1,); title(未受攻击的含水印图像); M1=a1; case 1 WImage2=a1; noise0=20*randn(size(WImage2); WImage2=WImage2+noise0; subplot(2,3,4); imshow(WImage2,); title(加入白噪声后图像); M1=WImage2; M_1=uint8(M1); imwrite(M_1,whitenoise.bmp,bmp); case 2 WImage3=a1; H=fspecial(gaussian,4,4,0.2); WImage3=imfilter(WImage3,H); subplot(2,3,4); imshow(WImage3,); title(高斯低通滤波后图像); M1=WImage3; M_1=uint8(M1); imwrite(M_1,gaussian.bmp,bmp); case 4 WImage4=a1; WImage4(1:64,1:512)=512; %WImage4(224:256,1:256)=256; %WImage4(1:256,224:256)=256; %WImage4(1:256,1:32)=256; WImage4cl=mat2gray(WImage4); figure(2); subplot(1,1,1); %subplot(2,3,4); imshow(WImage4cl); title(部分剪切后图像); figure(1); M1=WImage4cl; %M_1=uint8(M1); %imwrite(M_1,cutpart.bmp,bmp); case 3 WImage5=a1; WImage5=im2double(WImage5); cnum=10; dctm=dctmtx(8); P1=dctm; P2=dctm.; imageDCT=blkproc(WImage5,8,8,P1*x*P2,dctm,dctm.); DCTvar=im2col(imageDCT,8,8,distinct).; n=size(DCTvar,1); DCTvar=(sum(DCTvar.*DCTvar)-(sum(DCTvar)/n).2)/n; dum,order=sort(DCTvar); cnum=64-cnum; mask=ones(8,8); mask(order(1:cnum)=zeros(1,cnum); im88=zeros(9,9); im88(1:8,1:8)=mask; im128128=kron(im88(1:8,1:8),ones(16); dctm=dctmtx(8); P1=dctm.; P2=mask(1:8,1:8); P3=dctm; WImage5=blkproc(imageDCT,8,8,P1*(x.*P2)*P3,dctm.,mask(1:8,1:8),dctm); WImage5cl=mat2gray(WImage5); %figure(2); subplot(2,3,4); imshow(WImage5cl); title(经JPEG压缩后图像); %figure(1); M1=WImage5cl; case 5 WImage6=a1; WImage6=imrotate(WImage6,10,bilinear,crop); WImage6cl=mat2gray(WImage6); figure(2); subplot(1,1,1); imshow(WImage6cl); title(旋转10度后图像); figure(1); M1=WImage6cl; otherwise disp(你输入的是无效数字，图像未受攻击，将直接检测水印); subplot(2,3,4); imshow(a1,); title(未受攻击的含水印图像); M1=a1; end %6666666666666666666666666666666666666666 提取水印 % psnr_watermarked=M1; dca1=blkproc(M1,8,8,dct2); p=zeros(1,8); for i=1:dimI(1) for j=1:dimI(2) % j=1:32 x=(i-1)*8;y=(j-1)*8; p(1)=dca1(x+1,y+8); p(2)=dca1(x+2,y+7); p(3)=dca1(x+3,y+6); p(4)=dca1(x+4,y+5); p(5)=dca1(x+5,y+4); p(6)=dca1(x+6,y+3); p(7)=dca1(x+7,y+2); p(8)=dca1(x+8,y+1); %sd1=sum(sum(p.*k1)/sqrt(sum(sum(p.2); %sd2=sum(sum(p.*k2)/sqrt(sum(sum(p.2); %if sd1sd2 if corr2(p,k1)corr2(p,k2),warning off MATLAB:divideByZero; mark1(i,j)=1; else mark1(i,j)=0; end end end %mark=round(mark1); % 防止出现的那些小数，影响10-2进制的转换 subplot(2,3,5); imshow(mark1,),title(提取的水印图像); subplot(2,3,6); imwrite(mark1,getmark.bmp,bmp); imshow(mark),title(原嵌入水印比较); % time % disp(攻击与提取处理时间) attack_recover_time=cputime-start_time, % psnr % disp(载体图像与含水印图像峰值信噪比) PSNR=psnr(psnr_cover,psnr_watermarked,c,r), % Oringinal mark and mark test % disp(原水印图像与提取水印图像互相关系数) NC=nc(mark1,mark), 1、 数字水印的概念数字水印技术是指用信号处理的方法在数字化的多媒体数据中嵌入隐蔽的标记，这种标记通常是不可见的，只有通过专用的检测器或阅读器才能提取。通过这些隐藏在多媒体内容中的信息，可以确定内容的创建者，判断内容的真实完整性。数字水印技术需要增加鲁棒性要求，以对抗各种可能的攻击。2、 数字水印的基本特征隐藏性：数字产品加入数字水印后，应不会引起数字作品的明显质降，并不易被接收者察觉。安全可靠性：水印体制要商业应用，其算法必须公开。算法的安全性完全取决于密钥，而不是对算法进行保密。另外，数字水印应能对抗非法的探测和解码，面对非法的攻击也能以极低的差错率识别作品，同时数字水印应很难被他人复制和伪造。鲁棒性：能在多种无意或有意的信号处理过程后，仍能保持水印的完整性和鉴别的准确性。对数字水印来说，其隐藏水印的鲁棒性在实际应用中是有由两部分组成：（1）在整体数据出现失真后，其内嵌水印仍存在；（2）在数据失真后，水印探测算法仍能准确探测出水印的存在。计算复杂度：不同应用对水印嵌入算法和提取算法的计算复杂度有不同的要求。3、 数字水