Saving a grayscale image with scaled values

Question

Zinoviy 2022-11-8

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1846258-saving-a-grayscale-image-with-scaled-values

编辑： Jan 2022-11-8

采纳的回答： Jan

Hello, here is my problem:

I have a 196*196 image, and such matrix looks like that for example:

using imshow("img",[]) displays the image correctly because it's scaled, while not using it produces something like that:

I want to save the correct output that should looke like this:

But I can't find the correct parameters for imwrite...

Also I have to say that Im running it with a loop, and planning later on to do image processing, with batch processor and registration and etc.

6 个评论
显示 4更早的评论隐藏 4更早的评论

Jan 2022-11-8

在 MATLAB Online 中打开

folder = "directory"
for i = 1:100
   % Scale the image:
   img = fisp1(:,:,1,i);
   tmp = min(img(:));
   img = (img - tmp) / (max(img(:)) - tmp);
   % Save the file:
   name = fullfile(folder, sprintf('rep%d.png', i));
   imwrite(img, name);
end

Zinoviy 2022-11-8

Incredible, it worked.

Can you refer me to the explanation of this manual scaling?

Thanks!

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Jan 2022-11-8

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1846258-saving-a-grayscale-image-with-scaled-values#answer_1094238

编辑：Jan 2022-11-8

在 MATLAB Online 中打开

folder = "directory"
for i = 1:100
    % Scale the image:
    img = fisp1(:,:,1,i);
    tmp = min(img(:));
    img = (img - tmp) / (max(img(:)) - tmp);
    
    % Nicer scaling (Thanks Image Analyst):
    % normalize(img, 'range'), or: rescale(img), or: mat2gray(img)
    
    % Save the file:
    name = fullfile(folder, sprintf('rep%d.png', i));
    imwrite(img, name);
end

If the original values of the image have the range [a, b], so a is the minimal value and b maximal one.

Then img-a has range [0, b-a]. If you divide this by b-a, you get the range [0, 1]. This scales the interval [a,b] to [0,1] to get the maximum contrast.

2 个评论
显示无隐藏无

Image Analyst 2022-11-8

在 MATLAB Online 中打开

Instead of

tmp = min(img(:));
img = (img - tmp) / (max(img(:)) - tmp);

you can use the built in mat2gray

img = mat2gray(tmp);

or use rescale or normalize

Jan 2022-11-8

在 MATLAB Online 中打开

Image Analyst's suggestions are smart:

img = rand(1,8)
img = 1×8
    0.2261    0.6774    0.9945    0.3414    0.2208    0.1807    0.9315    0.6300
img2 = normalize(img, 'range')
img2 = 1×8
    0.0558    0.6103    1.0000    0.1975    0.0492         0    0.9227    0.5521
% Or:
img2 = rescale(img)
img2 = 1×8
    0.0558    0.6103    1.0000    0.1975    0.0492         0    0.9227    0.5521
% Or:
img2 = mat2gray(img)
img2 = 1×8
    0.0558    0.6103    1.0000    0.1975    0.0492         0    0.9227    0.5521
% Or less nice:
tmp = min(img(:));
img = (img - tmp) / (max(img(:)) - tmp)
img = 1×8
    0.0558    0.6103    1.0000    0.1975    0.0492         0    0.9227    0.5521