Not enough input argument
5 vues (au cours des 30 derniers jours)
Afficher commentaires plus anciens
function u =Blur_removal(f,A,mu,lambda,Niter,typeKernel)
% This function performs the minimization of
% u=arg min |D_x u|_1+|D_y u|_1+0.5*mu*||A°u-f||_2^2
%
% by Split Bregman Iteration
%
% f = blurry image
% A = convolution kernel
% mu = regularization parameter
% lambda = fidelity factor for the split variables
% Niter = maximum number of iterations
% typeKernel: 0 <=> A is considered in the spatial domain
% 1 <=> A is considered in the Fourier domain and must be of
% the same size of the image with frequency (0,0) at location
% (1,1)
[M,N]=size(f);
%Structures and constants initialization
f=double(f);
dx=zeros(M,N);
dy=zeros(M,N);
bx=zeros(M,N);
by=zeros(M,N);
u=f;
Z=zeros(M,N);
K=0;
%Fourier mask + initialization of Fourier constant quantities
if typeKernel==0
Mask=zeros(M,N);
[H,L]=size(A);
Mask([end+1-floor(H/2):end,1:ceil(H/2)],[end+1-floor(L/2):end,1:ceil(L/2)]) = A;
FMask=fft2(Mask);
else
FMask = A;
end
%Fourier Laplacian mask initialization
D = zeros(M,N);
D([end,1,2],[end,1,2]) = [0,1,0;1,-4,1;0,1,0];
FD=fft2(D);
%Fourier constant initialization
FW=((mu/lambda)*abs(FMask).^2-real(FD)).^-1;
FF=(mu/lambda)*conj(FMask).*fft2(f);
%Bregman iterations
err=norm(f(:),2);
tol=1e-3*err;
while ((err>tol) && (K<Niter)),
K=K+1;
tx=dx-bx;
ty=dy-by;
up=u;
%Update u
u=real(ifft2(FW.*(FF-fft2(tx-tx(:,[1,1:N-1])+ty-ty([1,1:M-1],:)))));
ux=u-u(:,[1,1:N-1]);
uy=u-u([1,1:M-1],:);
% dx
tmpx=ux+bx;
dx=sign(tmpx).*max(Z,abs(tmpx)-1/lambda);
% dy
tmpy=uy+by;
dy=sign(tmpy).*max(Z,abs(tmpy)-1/lambda);
% bx & by
bx=tmpx-dx;
by=tmpy-dy;
err=sum(sum((up-u).^2));
end
0 commentaires
Réponses (1)
KSSV
le 8 Août 2021
Don't run the code staright away using run/ f5/ green button. You need to provide the input variables and then call the function.
f = your input ;
A =
mu =
lambda =
Niter =
typeKernel =
u =Blur_removal(f,A,mu,lambda,Niter,typeKernel) ;
Define the above values and then call the function. The function has help, please read it.
0 commentaires
Voir également
Catégories
En savoir plus sur MATLAB Mobile dans Help Center et File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!