How can I reconstruct an RBG image?

I have the following image. Is there any way to reconstruct this image without all the noise and the blurring in it (with the same color of each fringe)?

In other words, I want to make these colored fringes on a black ground.

Answer 1

[Image Analyst]

How about this, using the attached data in answers.mat:

s = load('answers.mat')
red = repmat(s.red, [1699, 1]);
green = repmat(s.green, [1699, 1]);
blue = repmat(s.blue, [1699, 1]);
rgbImage = uint8(cat(3, red, green, blue));
imshow(rgbImage);
axis('on', 'image');

[Muhamed Sewidan]

Thank you for your answer, I think you have done a great job.

but there is one key component missing in the image, as you can see there is a shift for each fringe, the following image can describe the fringes and their shifts.

I had an idea that the Matlab can analyze each fringe colour and then redraw them again in their original path that includes the shits. Is it possible?

Thank you in advance.

[Muhamed Sewidan]

Excuse me but could you please tell how did you generate the answers file with the red, green and blue values?

[Image Analyst]

I took the average profile over a section of "good" rows in original image, then replicated it for the entire height of the image.

I assume the glitches were noise that you said you did not want, so I ignored them.

If you just want simple noise reduction, maybe you should try something like medfilt2() to do a median filter.

[Muhamed Sewidan]

How can I take the average profile (exuse me as I'm a beginner)? could you please tell me how to do it?

Acutally the shift in every fringe is the cornerstone of my image, therefore I want to neglect all the features of the image excpet every fringe and its shift. In other words, in my work i took images that have some problems such as noise, blurring and some unwanted objects (scratches on the slide of the specimen or dust particles). I had an idea that I can reconstruct the damaged image to create new one with the main features of the original image and without these problems.

Answer 2

[James Heselden]

I have tried to find you a solution which will work with any image you want, however I am only able to test this system on the single image you have provided. The following script has given me the attached output image. I hope this will help you with your project.

%Read in the image you want
I = imread('image.png');

%Extract and binarize the relevant colour channels
redChannel=imbinarize(I(:,:,1), 0.8);
greenChannel=imbinarize(I(:,:,2), 0.4);

%Filter the image
filtered = medfilt2(redChannel+greenChannel);

%Convert to binary mask
mask=imbinarize(filtered,0.3);

%Apply the mask to the colour channels
Ir = I(:,:,1); Ir(~mask)=0;
Ig = I(:,:,2); Ig(~mask)=0;
Ib = I(:,:,3); Ib(~mask)=0;

%Compile the output image
I_coloured(:,:,1) = Ir;
I_coloured(:,:,2) = Ig;
I_coloured(:,:,3) = Ib;

imagesc(I_coloured)