How to make a mask of a noiseless section of an image

Hi. I'm trying to detect and eliminate a blurred section of an image.

Right now I have found a way to detect a blurred section with Laplacian with the code below and reversed the colors with the last line (just for visualizing purposes), as seen on the image.

I realized that is the only section that is noiseless, since the rest of the image have a lot of noise. Is there a way to use that to just create a mask of the pure white (or black if is simpler) section? I was thinking of making an if section that checks for black pixels arround. It's not a problem if it masks a little extra on the edges.

Thanks for you help in advance.

clear all;
close all;
clc;

I = imread('a2.png');
H = zeros(size(I,1),size(I,2));

I_fft = fftshift(fft2(I));
for i = 1:size(I,1)
    for j = 1:size(I,2)
        H(i,j) = -((i-size(I,1)/2)^2+(j-size(I,2)/2)^2);
    end
end

I_lap = real(ifft2(ifftshift(I_fft.*H)));
I_lap_norm = I_lap/max(max(I_lap));

figure()
imshow(I_lap_norm)
imshow(imdilate(I_lap_norm,offsetstrel('ball',1,1)));

Answer 1

[Maneet Kaur Bagga]

Hi Fernando,

• As per my understanding to create a mask of the pure white section in the blurred image, a thresholding technique can be used. Since the pure white section corresponds to the least blurred region, it will have higher pixel values compared to the rest of the image.
• The "imbinarize" function can be used to convert the grayscale image "I_lap_norm" into a binary image based on the specified threshold value.
• Pixels with values above the threshold are set to 1 (white), and pixels below the threshold are set to 0 (black). The resulting binary image "mask" represents the pure white section of the blurred image.

% Threshold value to separate the pure white section
threshold = 0.9;

% Create a binary mask based on the threshold
mask = imbinarize(I_lap_norm, threshold);

% Display the mask
figure()
imshow(mask)

Please refer to the MATLAB documentation for further information on the "imbinarize" function:

https://www.mathworks.com/help/images/ref/imbinarize.html

Hope this helps!

Regards,

Maneet Bagga