数字水印中图像水印代码

%Coxwmk.m function [J, W] = CoxWmk(I, alpha, N, W) %COXWMK Insert a random watermark. % [J, W] = COXWMK(I) watermarks the grayscale picture I using % Cox's scheme [2] with default parameters. J is the watermarked % picture and W is the watermark. % The watermark W is drawn from a Normal random process N(0,1) % if it is not given by the user as last parameter. % % [J, W] = COXWMK(I, Alpha, N) watermarks the grayscale picture I % using modifying the Nth largest DCT coefficients v_i in the % following way: v_i := v_i * (1 + Alpha * w_i) where w_i % are the components of the watermark. % % See: COXDETECT % This is a modified version of Fabien Petitcolas' implementation % of Cox's technique by Gabriela Delfino and Fabian Martinez % % Fabien Petitcolas' website: http://www.cl.cam.ac.uk/~fapp2 if (nargin == 1) N = 1000; % alpha = 0.1; alpha = 0.01; % We modify the value of alpha to achieve better visual quality W = randn(1,N); end if (nargin == 2) N = 1000; W = randn(1,N); end if (nargin == 3) W = randn(1,N); end sI = size(I); if ((sI(1) * sI(2)) < N) error('Image too small or too many coefficients.'); end % CoxWmk also works with color images % In that case we watermark the Y component of the & imagen in YIQ representation. copiaI=I; % If the imagen is a color image we just watermark the Y plane if (isrgb(I)) IenYIQ=RGBtoYIQ(I); I=IenYIQ(:,:,1); end % Compute the DCT of the image DCTI = dct2(I); % Find the N largest coefficients in the DCT matrix % Better if the extraction of the N largest was done % at the same time than the computation of the DCT... Index = FindNLargest(abs(DCTI), N); % Modify these coefficients for i = 1:N DCTI(Index(1,i),Index(2,i)) = DCTI(Index(1,i),Index(2,i)) * (1 + alpha * W(i)); end % Simply take the inverse DCT of the modifyied matrix % J = idct2(DCTI); IconMarca = idct2(DCTI); % We save the inverse DCT results in an auxiliar matrix % If the image was a color image we convert it to RGB again if (isrgb(copiaI)) JenYIQ=IenYIQ; JenYIQ(:,:,1)=IconMarca; J=YIQtoRGB(JenYIQ); else J=IconMarca; end % We change the values above 255 to 255 in order to avoid posible % distortions caused by the double representation of MatLab J=abs(J); J=cambiarValores(J,255,255); J=uint8(J);