Inpainting image, it just wont work, i have followed the equation provided

%%I am having writting an inpainting code and having an issue with my code, where i am trying to convert my missing pixels to i and j co-ordinates



close all
% clear; Please remember to uncomment this when you submit your code as an assignment submission.

% Read in the picture
original = double(imread('greece.tif'));

% Read in the forcing function
load forcing;

% Read in the corrupted picture which contains holes
load badpicture;

% Read in an indicator picture which is 1 where the
% pixels are missing in badicture
mask = double(imread('badpixels.tif'));


% Initialise iterations &variables here
restored = badpic; 
restored2=badpic;

total_iterations=2000;
err1 = zeros(1, total_iterations);
err2 = zeros(1, total_iterations);
alpha=1;
forcing=load('forcing.mat');
f=forcing.f;


% This displays the original picture in Figure 1
figure(1);
image(original);
title('Original');
colormap(gray(256));

% Display the corrupted picture in Figure 2
figure(2); 
image(badpic);
title('Corrupted Image');
colormap(gray(256));

    

finder = find(mask ~= 0); % This stores all the locations in vectors i and j (row and column indices for each missing pixel)
spaceD=[720,1280];
[i,j]=ind2sub(spaceD, finder);

for iteration = 1 : total_iterations, 
  for n=1:61440
        total = badpic(i(n) - 1, j(n)) + badpic(i(n) + 1, j(n))...
        + badpic(i(n), j(n) + 1) + badpic(i(n), j(n) - 1);
        update = badpic(i(n), j(n));
        update = update + alpha * (total - 4 * (badpic(i(n), j(n))))/4;
        restored(i(n), j(n)) = update;
    end
    
    err1(i)= std(badpic(finder) - original(finder));

end;

% Display the restored image in Figure 3 (This does NOT use the forcing function)
figure(3);
image(restored); 
title('Restored Image');
colormap(gray(256));

% Repeat the restoration, again starting from the badpicture, but using the forcing function in update
for iteration = 1 : total_iterations,
  for n=1:61440
   total = badpic(i(n) - 1, j(n)) + badpic(i(n) + 1, j(n))...
        + badpic(i(n), j(n) + 1) + badpic(i(n), j(n) - 1)- f(i(n), j(n));
        update = badpic(i(n), j(n));
        update = update + alpha * (total - 4 * (badpic(i(n), j(n)))) / 4;
        restored2(i(n), j(n)) = update;
end
err2(i)= std(badpic(finder) - original(finder));
end;

% Display your restored image with forcing function as Figure 4
figure(4);
image(restored2); 
title('Restored Image with forcing function');
colormap(gray(256));

% Plot two error vectors versus iteration
figure(5);
plot(total_iterations,err1, 'r-', total_iterations, err2, 'b-', 'linewidth', 3);
legend('No forcing function', 'With forcing function');
xlabel('Iteration', 'fontsize', 20);
ylabel('Std Error', 'fontsize', 20);





 

[John D'Errico]

I'd just download and use my inpaint_nans from the file exchange. Are you really wedded to writing it yourself?

Answer 1

[Image Analyst]

Why not simply use regionfill() the built-in function for doing it?